Unraveling the Carcinogenic Mechanisms of Food contaminants through Network Toxicology, Machine Learning, and Molecular Docking

Unraveling the Carcinogenic Mechanisms of Food contaminants through Network Toxicology, Machine Learning, and Molecular Docking | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Unraveling the Carcinogenic Mechanisms of Food contaminants through Network Toxicology, Machine Learning, and Molecular Docking Bangsheng Chen, Maomao Li, Yi Gu, Wenzhu Lou, Shuaishuai Huang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6625099/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Food contamination is a significant global health threat, with carcinogenic potential, yet the molecular pathways linking contaminants to cancer remain poorly understood. This study aimed to identify key molecular targets mediating the carcinogenic effects of food contaminants. We utilized multiple online databases to identify target genes associated with nine prevalent dietary contaminants (Glyphosate, Perfluorooctane sulfonate, Nitrosamines, Pentabromodiphenyl ethers, Methylmercury, Dioxins, Acrylamide, Pyrrolizidine Alkaloids, and Aflatoxin) and pan-cancer. Protein-protein interaction (PPI) analysis and visualization were conducted on intersecting genes, followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analyses to uncover potential mechanisms. We focused on breast (BRCA), prostate (PRAD), and colon (COAD) carcinomas due to their significant pathway associations. Hub genes were prioritized using an integrative strategy combining topological algorithms in Cytoscape (Centiscape, MCODE, and cytohubba's MCC), machine learning validation, and Weighted Gene Co-expression Network Analysis (WGCNA). Molecular docking simulations were performed to examine interactions between contaminants and hub genes. We identified 69 pan-cancer-intersected targets. Comprehensive enrichment analyses revealed significant cancer-associated pathways. Hub gene prioritization identified JUN in BRCA, CDC42 in COAD, and MAPK14 in PRAD as critical regulatory targets. Validation using The Cancer Genome Atlas (TCGA) data confirmed statistically significant differential expression patterns (p < 0.05) for these targets across respective malignancies. Gene Set Enrichment Analysis (GSEA) delineated pathway activation profiles consistent with tumor progression mechanisms. Molecular docking simulations demonstrated robust binding affinities (binding energy ≤-5.0 kcal/mol) between contaminants and structural domains of identified hub targets, suggesting direct mechanistic interactions. Our study elucidates the molecular mechanisms underlying dietary carcinogens, identifies potential therapeutic targets, and highlights the need for enhanced food safety policies. This integrative approach combining molecular and clinical insights may inform precision public health interventions. Biological sciences/Cancer Earth and environmental sciences/Environmental social sciences Health sciences/Health care Food contamination Cancer Network Toxicology Machine learning Molecular docking Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 Figure 12 Figure 13 1. Introduction Cancer remains a leading cause of mortality worldwide, accounting for nearly 10 million deaths annually, with projections suggesting a 50% increase in incidence by 2050 1 . Although advancements in oncology have shed light on the genetic and molecular drivers of carcinogenesis, approximately 20–30% of cancers are linked to modifiable environmental factors, including chemical exposures 2 , 3 . Despite increasing awareness, public understanding of pervasive environmental carcinogens - such as air pollutants, industrial chemicals, and food contaminants - remains limited. This gap in knowledge perpetuates preventable exposures, contributing to the rising incidence of cancer and highlighting the urgent need to identify and mitigate these hazards. Among environmental carcinogens, food contaminants constitute a significant yet often underestimated threat 4 , 5 . Compounds such as glyphosate (herbicide residues), perfluorooctane sulfonate (PFOS, a persistent organic pollutant), nitrosamines (byproducts of processed meats), and aflatoxins (mycotoxins found in improperly stored crops) are prevalent throughout global food supply chains 6 – 9 . Epidemiological research has established direct associations between these substances and various malignancies: for instance, aflatoxin B1 is implicated in the induction of hepatocellular carcinoma through mechanisms involving apoptosis and autophagy, among others 10 . Furthermore, exposure to PFOS has been linked to an increased risk of thyroid and lung cancers 11 . Acrylamide, which forms during high-temperature cooking processes, is classified by the International Agency for Research on Cancer (IARC) as a Group 2A carcinogen due to its genotoxic properties 12 . Despite the substantial body of evidence supporting these associations, regulatory measures have not kept pace with scientific advancements, highlighting the urgent need for mechanistic insights to guide policy development and public health strategies. Traditional toxicology frequently encounters limitations in deciphering the multi - target mechanisms of foodborne carcinogens. Network toxicology addresses this limitation by integrating systems biology and computational modeling to map chemical-gene-disease interactions, thereby identifying central hubs within carcinogenic pathways 13 . For instance, Li et al. (2025) employed PPI networks and machine learning to identify key genes (HDAC6, CDK1, DNMT1, NOS3, DPP4) that mediate the effects of air pollutants on prostate cancer 14 . Similarly, He et al. (2024) applied network toxicology, protein interactions, and molecular docking to elucidate the interactions between plasticizers and key proteins implicated in breast cancer 15 . These studies effectively address the "needle-in-a-haystack" challenge of target prioritization. Additionally, Xu et al. (2024) utilized differential expression analysis, WGCNA, database mining, machine learning, and molecular docking to elucidate the therapeutic mechanisms of Hypericum perforatum in major depressive disorder (MDD) 16 . Their integrated approach identified key targets and pathways, highlighting immune-related effects and providing a robust framework to translate high-dimensional data into actionable insights for disease prevention. These methodologies collectively bridge the gap between chemical exposure and phenotypic outcomes. Molecular docking simulations offer atomic level insights into carcinogen-target interactions, complementing network based predictions 17 . For instance, the study by Manal A. Abbas et al. (2024) demonstrated a high-affinity binding between bisphenol A and estrogen receptor α, shedding light on its role in the progression of breast cancer 18 . Similarly, Babak Arjmand et al. (2019) utilized molecular docking to align ligand poses, rank active compounds, and propose potential anticancer agents 19 . By integrating docking results with network topology, we can validate hub targets and refine mechanistic models, thus providing a robust toolkit for risk assessment. This study introduces a pioneering multi-modal framework that combines network toxicology, machine learning, WGCNA, and molecular docking to unravel foodborne carcinogenesis. We systematically examined nine high-risk contaminants, identified pan-cancer targets and pathway modules, and validated hub genes (JUN, CDC42, MAPK14) in breast, colon, and prostate cancers using TCGA cohorts. Our approach not only delineates novel mechanisms but also provides actionable targets for therapeutic development. By linking molecular signatures to clinical outcomes, this work underscores the imperative of food safety reforms and precision public health strategies to mitigate the global cancer burden. 2. Methods 2.1. Identification of physicochemical properties and toxicity of food contaminatants. We obtained the chemical structures and corresponding molecular details of nine food contaminants from the PubChem database 20 ( https://pubchem.ncbi.nlm.nih.gov ). The carcinogenic properties of these pollutants were subsequently assessed using the ADMETLAB 3.0 platform 21 ( https://admetlab3.scbdd.com ) and the ProTox3 database 22 ( https://tox.charite.de/protox3 ) as evidenced in Table 1 . 2.2. Collection of food contaminants target genes The potential human target genes associated with the nine food contaminants were sourced from the STITCH database 23 ( http://stitch.embl.de/cgi/ ), the Swiss Target Prediction database 24 ( http://www.swisstargetprediction.ch/ ), the SEA database 25 ( https://sea.bkslab.org/ ), and the TargetNet database ( http://targetnet.scbdd.com ). The gene data obtainedfrom these four databases were consolidated, duplicates were eliminated, and a definitive set of target genes for the food contaminants was established. 2.3. Collection of cancer-related genes Cancer-related genes were sourced from two databases: OMIM 26 ( https://omim.org/ ) and GeneCards 27 ( https://www.genecards.org/ ). The genes were selected based on specific criteria: a "score" threshold equivalent to the median value from GeneCards was applied, and only genes exceeding this median score were chosen. Following the consolidation of gene lists from both databases and the subsequent removal of duplicate entries, a comprehensive set of cancer-associated genes was established. 2.4. The intersection of food contaminants targets and cancerrelated targets The intersection analysis between contaminant targets and cancer-related targets was conducted using the Venn diagram tool available on the web ( http://www.bioinformatics.com.cn/ ). Both the contaminant target library and the cancer target library were uploaded to this platform, and the results were retrieved upon completion of the analysis. 2.5. Construction of protein - protein interaction (PPI) networks We constructed the PPI network utilizing the STRING database 28 ( https://cn.string-db.org/)wit h overlapping targets identified from the Venn diagram. The Table 1 The physicochemical properties and toxicity assessment of compounds CAS Compound Term Molecular Formula Molecular Weight(g/mol) LogP LC 50 DM LC 50 FM IGC 50 BCF Carcinogenicity In ProTox 3.0 Carcinogenicity In ADMETLab3 1071-83-6 Glyphosate C 3 H 8 NO 5 P 169.07 -3.419 3.259 3.032 2.439 0.15 0.58 0.896 1763-23-1 Perfluorooctane sulfonate C 8 HF 17 O 3 S 500.13 1.817 6.341 5.744 4.039 3.178 0.68 0.448 81407-93-4 Nitrosamines H 2 N 2 O 46.029 -0.617 2.769 1.826 1.603 0.13 0.81 0.998 1163-19-5 Pentabromodiphenyl ethers C 12 Br 10 O 959.2 6.977 7.666 9.13 6.161 0.719 0.85 0.905 16056-34-1 Methyl mercury CH 3 Hg 215.63 1.166 4.269 3.858 3.169 1.036 0.68 0.984 1746-01-6 Dioxins C 12 H 4 Cl 4 O 2 322 6.673 5.861 6.305 5.341 4.383 0.89 0.73 79-06-1 Acrylamide C 3 H 5 NO 71.08 -0.513 2.779 2.861 2.307 0.522 0.89 0.714 81340-07-0 Pyrrolizidine alkaloid C 18 H 23 NO 5 333.4 0.63 4.953 4.054 3.318 0.56 0.61 0.58 1165-39-5 Aflatoxin C 17 H 12 O 7 328.27 1.654 6.146 4.973 3.809 1.296 0.6 0.992 “CAS”: Chemical Abstracts Service; “LogP”: The logarithm of the n-octanol/water distribution coefficients at pH = 7.4; “BCF”:Bioconcentration factors; “LC 50 DM”: 50% lethal concentrationin the daphnia magna after 48 h; “LC 50 FM”: 50% lethal concentration in the fathead minnow after 96 h; “IGC 50 ”: 50% growth inhibition concentration in the tetrahymena pyriformis. species was specified as Homo sapiens, and an interaction score of ≥ 0.9 was applied to generate the PPI network diagram. The data obtained from STRING were imported into Cytoscape 3.10.3 29 for network visualization and analysis, facilitating the computation of topological properties and the creation of a PPI network diagram. Core targets were selected based on nodes exceeding the median values for betweenness centrality, closeness centrality, and degree. The MCODE plugin was utilized to identify the most significant subclusters of interacting nodes, and the CytoHubba plugin was employed to predict the top 10 significant genes using the maximal clique centrality (MCC) algorithm. Hub genes were ascertained by intersecting the targets of the MCODE significant module with those predicted by CytoHubba. 2.6. GO and KEGG pathway analysis GO terms, encompassing biological processes, molecular functions, and cellular components, along with KEGG pathway analyses, were conducted utilizing the DAVID database 30 ( https://david.ncifcrf.gov/ ) to explore the pathways associated with cancer-related interactome targets. This analysis was designed to reveal and underscore the significant signaling pathways engaged in these biological processes. Visual analysis was implemented to efficiently interpret and display the outcomes of the GO and KEGG analyses, facilitated by the visualization platform provided at https://www.bioinformatics.com.cn . 2.7. Screening of hub targets using machine learning algorithms We employed four distinct machine learning algorithms for target screening 31 . Lasso regression analysis, SVM-R feature extraction (SVM-RFE), random forest (RandomForest), and Boruta feature selection were applied to the genes within the key subnetwork, using a random seed of “123” in RStudio (version 4.4.2). Genes identified at the intersection of these analyses were deemed as potential targets for toxicities associated with cancers. Furthermore, a nomogram depicting the five-year survival probability of cancer patients was constructed using the Rms package, demonstrating the clinical utility of these interactome targets. 2.8. Identification of potential target genes through Weighted Gene Co-expression Network Analysis (WGCNA) For the WGCNA analysis 32 , we sourced cancerous and normal tissues from the UCSC Xena database. We confirmed the soft threshold using the pickSoftThreshold function from the WGCNA package in R software. After establishing a scale-free network based on this soft threshold, a topology matrix and hierarchical clustering were applied. Subsequently, Eigengenes were identified for each module. Based on these Eigengenes, hierarchical clustering was performed once the correlation between modules was established. The module eigengene (ME), considered representative of the gene expression profiles, was calculated to identify modules associated with clinical traits. To pinpoint the most tumor-related modules, we conducted Module-Trait Relationships calculations for each module. Eigengene Adjacency Heatmaps were generated to illustrate module relationships. 2.9. Characterization of key targets for expression, survival probability, and gene set enrichment analysis (GSEA) The expression profiles of the identified key targets were correlated and visualized using the corrplot package within RStudio (version 4.4.3). The expression levels of each critical gene were assessed through the Wilcoxon rank sum test. Subsequently, GSEA was conducted for each key gene to gain further insights into the functions of the pathways enriched. 2.10. Molecular docking We integrated data from the Protein Data Bank (PDB) and UniProt databases 33 to acquire crystal structures of key targets, preprocessing them with AutoDock Vina version v1.2.7 34 . The PDB files utilized in this study were JUN (5FV8), MAPK14 (6SFO), and CDC42 (2QRZ). Detailed information on these protein structures is presented in Supplementary Table 1. The 3D structures of the small molecules were sourced from the PubChem database ( https://pubchem.ncbi.nlm.nih.gov/ ) and stored in SDF format. The CAS numbers of the core toxic targets were employed to fetch their 3D structures from the PDB database ( http://www.rcsb.org/ ), with Homo sapiens specified as the species and resolution refined between 0–3.0 Å. Subsequently, we executed AutoDock for molecular docking of macromolecules and ligands using default parameters. Each docking pair was performed nine times, and the lowest binding affinity was documented. The docking models were visualized, and the binding energy was illustrated in a heat map using the pheatmap package within RStudio (version 4.4.3). 3. Results 3.1. Physicochemical and Toxicological Assessment of Contaminants A thorough evaluation of the physicochemical properties and toxicological profiles of nine representative contaminants was conducted utilizing systematic ADMET and ProTox analyses, as illustrated in Fig. S1. A significant positive correlation was observed between molecular weight and lipophilicity across the contaminants under study (Fig. S1A). Furthermore, the environmental toxicity of the contaminants was assessed through lethal concentrations (LC 50 ) against Tetrahymena thermophila, Pimephales promelas, and Daphnia magna, revealing consistent variations in bioconcentration factors among the nine contaminants (Fig. S1B). Notably, the carcinogenic potential of these contaminants was found to be greater than 0.5 in both ProTox 3.0 and ADMETlab 3 platforms (Fig. S1C). These findings are consistent with existing clinical knowledge regarding the toxicity of the selected contaminants, confirming their particularly potent carcinogenic properties. 3.2. Identification of food contaminant-pan-cancer core targets and PPI network construction We performed a comprehensive screening of 450 food contaminant targets from databases including STITCH, Swiss Target Prediction, SEA, and TargetNet, identifying 1852 pan-cancer targets from GeneCards and OMIM. A Venn diagram revealed 69 potential targets associated with contaminant-induced pan-cancer (Fig. 1A). Using STRING, we constructed PPI networks and refined them in Cytoscape with criteria like Closeness unDir > 0.007993356128408048, Betweenness unDir > 63.617647058823486, and Degree unDir > 17.676470588235293, identifying 14 core targets (Figs. 1B, C). We then performed GO and KEGG enrichment analyses using DAVID, focusing on Homo sapiens. This generated 130 KEGG entries and 276 GO entries, including 167 biological processes, 22 cellular components, and 87 molecular functions. The top 10 GO entries with the highest gene ratios were highlighted in a bar chart (Fig. 1D), showing key processes like positive regulation of gene expression, chromatin remodeling, and positive regulation of transcription by RNA polymerase II. The top 20 KEGG entries were plotted (Figs. 1E,F), highlighting cancer-related pathways like pathways in cancer, PI3K-Akt signaling pathway, Proteoglycans in cancer.We selected three cancers for further analysis. Finally, we visualized the contaminant-target-pathway network in Cytoscape (Fig. 1G), with 9 contaminants (green), 20 pathways (orange), related target genes (blue), and 14 potential hub genes (red). This diagram clearly shows the complex relationships between contaminants, pathways, and target genes. 3.3. The effect of food comtaminants on BRCA We investigated the impact of food contaminants on breast cancer by integrating 15,632 targets from GeneCards and OMIM databases with 450 food contaminant targets, identifying 369 potential targets associated with food contaminant-induced breast cancer (Fig. 2A). Using the STRING database, we constructed PPI networks for these targets and imported them into Cytoscape for analysis. We applied criteria including Closeness unDir > 0.0010247338113175926, Betweenness unDir > 758.5291828793742, and Degree unDir > 6.64591439688716 to select 49 core targets (Fig. 2B). The MCODE plugin in Cytoscape classified disease targets associated with breast cancer (Fig. 2C), and the CytoHubba plugin identified hub genes, revealing 5 subnetworks and the top 10 targets with the highest degree ranking (Fig. 2E). GO and KEGG enrichment analysis using the DAVID database yielded 161 KEGG entries and 634 GO entries, including 441 biological processes (BP), 51 cellular components (CC), and 142 molecular functions (MF). The top 10 entries in BP, CC, and MF were selected for GO enrichment analysis, and the top 20 entries for KEGG analysis. The BP category was associated with transcription regulation by RNA polymerase II, positive gene expression regulation, apoptosis regulation, and chromatin remodeling. The CC category related to cytoplasm, cytosol, and nucleus, while MF concerned protein binding, enzyme binding, and identical protein binding (Fig. 2F). KEGG pathway analysis indicated significant involvement in cancer pathways, including pathways in cancer, proteoglycans in cancer, and chemical carcinogenesis-receptor activation (Fig. 2D). These insights shed light on the molecular mechanisms of food contaminant-induced breast cancer. Machine learning algorithms were employed to identify key targets associated with breast cancer induced by food contaminants. Feature screening was executed by constructing LASSO regression models, identifying 31 genes (Figs. 3A, B). The SVM-RFE algorithm was also utilized to assess signature genes, resulting in the identification of 10 optimal signature genes (Fig. 3E). Random forest classification and Boruta feature importance algorithms were applied to select potential signature genes (Figs. 3C, D). By intersecting the signature genes obtained from these four models, six genes (PPARG, TLR4, EGFR, CASP3, CDK5, JUN) were pinpointed as key targets for subsequent breast cancer analysis (Fig. 3F). A nomogram depicting the five-year survival probability for breast cancer was crafted using the Rms package, underscoring the clinical utility of these targets (Fig. 3G). WGCNA was performed using the TCGA-BRCA dataset to identify key gene modules. Samples were grouped based on Pearson correlation, with a soft threshold power set at 10 for optimal clustering (Figs. S2A, B). The dynamic tree cut package identified 17 distinct modules (Fig. S2C). Module-trait relationships were established to assess correlations with tumor occurrence. Significant positive correlations with tumor development were observed for the green-yellow (coefficient 0.46) and grey modules (coefficient 0.5), while the brown module showed a significant positive correlation with normal conditions (coefficient 0.66) (Fig. S2D). Eigengene adjacency heatmaps depicting these relationships are illustrated in Fig. S2E. This analysis provides potential targets into the molecular distinctions between tumor and normal tissue within BRCA. Interacting with hub genes selected by machine learning, two targets, JUN and EGFR, were identified. RNA sequencing (RNAseq) data in the Transcripts Per Million (TPM) format from the UCSC Xena database were processed by the Toil process. Normal tissue data of BRCA and GTEX from TCGA were extracted, consisting of 179 normal and 1,099 tumor samples. RNAseq data were transformed into log2 format, and expression comparisons were performed (Fig. 4A). Wilcoxon rank sum test revealed that JUN and EGFR expression was significantly lower in tumor samples than in normal samples. Pairing analysis between normal and tumor tissues from the same patient showed similar results (Figs. 4B-E). Survival analysis indicated that BRCA patients with higher quartile JUN expression had longer survival than those with lower quartile JUN expression (Figs. 4F,G). GSEA functional analysis of JUN identified multiple associated pathways, including apical junction, epithelial-mesenchymal transition, hypoxia, e2f targets, and g2m checkpoint (Figs. 4H,I). Molecular docking analysis was performed with JUN and 9 food contaminants. The lowest binding energies between these contaminants and the JUN protein were visualized. The contaminants interacted with key amino acid residues of JUN, including ARG-21, ARG-16, GLN-12, ARG-28, GLU-7, and GLU-15. A binding energy of -6.93 kcal/mol indicated stable binding affinity between JUN and the contaminants (Fig. 5). 3.4. The effect of food comtaminants on PRAD We investigated the impact of food contaminants on prostate cancer by integrating 9,629 targets from GeneCards and OMIM databases with 450 food contaminant targets, identifying 323 potential targets associated with food contaminant-induced prostate cancer (Fig. 6A). Using the STRING database, we constructed protein-protein interaction (PPI) networks for these targets and imported them into Cytoscape for analysis. We applied criteria including Closeness unDir > 0.0011772220504692277, Betweenness unDir > 650.0086206896538, and Degree unDir > 6.793103448275862 to select 44 core targets (Fig. 6B). The MCODE plugin in Cytoscape classified disease targets associated with prostate cancer (Fig. 6C), and the CytoHubba plugin identified hub genes, revealing 3 subnetworks and the top 10 targets with the highest degree ranking (Fig. 6E). GO and KEGG enrichment analysis using the DAVID database yielded 163 KEGG entries and 606 GO entries, including 415 biological processes (BP), 53 cellular components (CC), and 138 molecular functions (MF). The top 10 entries in BP, CC, and MF were selected for GO enrichment analysis, and the top 20 entries for KEGG analysis. The BP category was associated with regulation of transcription by RNA polymerase II, regulation of gene expression, regulation of apoptotic process, and regulation of DNA-templated transcription. The CC category related to cytoplasm, nucleus, and cytosol, while MF concerned enzyme binding, identical protein binding, and protein kinase activity (Fig. 6F). KEGG pathway analysis indicated significant involvement in cancer pathways, including pathways in cancer, proteoglycans in cancer, and chemical carcinogenesis-receptor activation (Fig. 6D). These insights shed light on the molecular mechanisms of food contaminant-induced prostate cancer. Machine learning algorithms were employed to identify key targets associated with prostatecancer induced by food contaminants. Feature screening was executed by constructing LASSO regression models, identifying 22 genes (Figs. 7A, B). The SVM-RFE algorithm was also utilized to assess signature genes, resulting in the identification of 30 optimal signature genes (Fig. 7E). Random forest classification and Boruta feature importance algorithms were applied to select potential signature genes (Figs. 7C, D). By intersecting the signature genes obtained from these four models, fifteen genes (APP, CDC42, CDK5, CYP19A1, EGFR, ESR1, ESR2, HDAC1, HSP90AA1, HSP90AB1, MAPK14, MMP9, PPARG, PRKACA, PTGS2) were pinpointed as key targets for subsequent prostate cancer analysis (Fig. 7F). A nomogram depicting the five-year survival probability for prostate cancer was crafted using the Rms package, underscoring the clinical utility of these targets (Fig. 7G). WGCNA was performed using the TCGA-PRAD dataset to identify key gene modules. Samples were grouped based on Pearson correlation, with a soft threshold power set at 10 for optimal clustering (Figs. S3A, B). The dynamic tree cut package identified 14 distinct modules (Fig. S3C). Module-trait relationships were established to assess correlations with tumor occurrence. Significant positive correlations with tumor development were observed for the black (coefficient 0.32) and turquoise modules (coefficient 0.32), while the pink (coefficient 0.41) and magenta modules (coefficient 0.45) showed a significant positive correlation with normal conditions (Fig. S3D). Eigengene adjacency heatmaps depicting these relationships are illustrated in Fig. S3E. This analysis provides potential targets into the molecular distinctions between tumor and normal tissue within PRAD. Interacting with hub genes selected by machine learning, two targets, MAPK14,and CDC42were identified. RNAseq data in the TPM format from the UCSC Xena database were processed using the Toil process. Data from normal tissues of PRAD and GTEX from The TCGA were extracted, encompassing 100 normal and 496 tumor samples. RNAseq data were normalized to the log2 scale, facilitating expression comparisons (Fig. 8A). The Wilcoxon rank sum test indicated significantly reduced expression of MAPK14 and CDC42 in tumor samples compared to normal samples. Pairwise analysis between matched normal and tumor tissues from the same patient corroborated these findings for MAPK14, though no significant difference was observed for CDC14 (Figs. 8B-E). Survival analysis revealed that patients with lower quartile MAPK14 expression in PRAD exhibited longer survival times than those with higher MAPK14 expression (Figs. 8F, G). GSEA of MAPK14 identified multiple associated pathways, including allograft rejection, epithelial-mesenchymal transition, inflammatory response, adipogenesis, DNA repair, and coagulation (Figs. 8H, I). Molecular docking analysis was performed with MAPK14 and 9 food contaminants. The lowest binding energies between these contaminants and the MAPK14 protein were visualized. The contaminants interacted with key amino acid residues of MAPK14, including MET-109, HIS-107, SER-251, SER-252, LYS-249, TYR-132, ARG-136, ASN-82, LEU-151, ASP-150, THR-185, PRO-191, and LEU-195. A binding energy of -9.63 kcal/mol indicated stable binding affinity between MAPK14 and the contaminants (Fig. 9). 3.5. The effect of food comtaminants on COAD We investigated the impact of food contaminants on colorectal cancer by integrating 11,550 targets from GeneCards and OMIM databases with 450 food contaminant targets, identifying 325 potential targets associated with food contaminant-induced colorectal cancer (Fig. 10A). Using the STRING database, we constructed PPI networks for these targets and imported them into Cytoscape for analysis. We applied criteria including Closeness unDir > 0.0012390264023192084, Betweenness unDir > 611.297777777777, and Degree unDir > 6.844444444444444 to select 48 core targets (Fig. 10B). The MCODE plugin in Cytoscape classified disease targets associated with colorectal cancer (Fig. 10C), and the CytoHubba plugin identified hub genes, revealing 2 subnetworks and the top 10 targets with the highest degree ranking (Fig. 10E). GO and KEGG enrichment analysis using the DAVID database yielded 162 KEGG entries and 630 GO entries, including 434 biological processes (BP), 55 cellular components (CC), and 141 molecular functions (MF). The top 10 entries in BP, CC, and MF were selected for GO enrichment analysis, and the top 20 entries for KEGG analysis. The BP category was associated with regulation of transcription by RNA polymerase II, regulation of gene expression, regulation of apoptotic process, and regulation of DNA-templated transcription. The CC category related to cytoplasm, nucleus, and cytosol, while MF concerned protein binding, identical protein binding, enzyme binding, and ATP binding (Fig. 10F). KEGG pathway analysis indicated significant involvement in cancer pathways, including pathways in cancer, proteoglycans in cancer, and chemical carcinogenesis - reactive oxygen species (Fig. 10D). These insights shed light on the molecular mechanisms of food contaminant-induced colorectal cancer. Machine learning algorithms were employed to identify key targets associated with colorectal cancer induced by food contaminants. Feature screening was executed by constructing LASSO regression models, identifying 14 genes (Figs. 11A, B). The SVM-RFE algorithm was also utilized to assess signature genes, resulting in the identification of 30 optimal signature genes (Fig. 11E). Random forest classification and Boruta feature importance algorithms were applied to select potential signature genes (Figs. 11C, D). By intersecting the signature genes obtained from these four models, twelve genes (ABL1, AHR, CDC42, CYP19A1, ESR2, GSK3B, HSP90AB1, MMP9, NR3C1, PRKACA, PSEN1, RIPK2) were pinpointed as key targets for subsequent colorectal cancer analysis (Fig. 11F). A nomogram depicting the five-year survival probability for colorectal cancer was crafted using the Rms package, underscoring the clinical utility of these targets (Fig. 11G). WGCNA was performed using the TCGA-COAD dataset to identify key gene modules. Samples were grouped based on Pearson correlation, with a soft threshold power set at 10 for optimal clustering (Figs. S4A, B). The dynamic tree cut package identified 21 distinct modules (Fig. S4C). Module-trait relationships were established to assess correlations with tumor occurrence. Significant positive correlations with tumor development were observed for the lightcyan (coefficient 0.57) and magenta modules (coefficient 0.38), while the yellow (coefficient 0.78) and turquoise modules (coefficient 0.72) showed a significant positive correlation with normal conditions (Fig. S4D). Eigengene adjacency heatmaps depicting these relationships are illustrated in Fig. S4E. This analysis provides potential targets into the molecular distinctions between tumor and normal tissue within COAD. Interacting with hub genes selected by machine learning, two targets, ESR2, and CDC42 were identified. RNAseq data in the TPM format from the UCSC Xena database were processed using the Toil process. Data from normal tissues of COAD and GTEX from The TCGA were extracted, encompassing 308 normal and 290 tumor samples. RNAseq data were normalized to the log2 scale, facilitating expression comparisons (Fig. 12A). The Wilcoxon rank sum test indicated significantly reduced expression of CDC42 and ESR2 in tumor samples compared to normal samples. Pairing analysis between normal and tumor tissues from the same patient showed similar results (Figs. 12B-E). Survival analysis revealed that patients with lower quartile CDC42 expression in COAD exhibited lower survival probability than those with higher CDC42 expression, though no significant difference was observed for ESR2 (Figs. 12F, G). GSEA of CDC42identified multiple associated pathways, including allograft_rejection, epithelial-mesenchymal transition, inflammatory response, interferon_gamma_response, and myc_targets_v2 (Figs. 12H, I). Molecular docking analysis was performed with CDC42 and 9 food contaminants. The lowest binding energies between these contaminants and the CDC42 protein were visualized. The contaminants interacted with key amino acid residues of CDC42, including THR-17, THR-58, ASP-57, ASP-11, ASP-176, LYS-16, ALA-13, ALA-176, ARG-186, GLY-15, GLU-178, and GLY-164. A binding energy of -7.47 kcal/mol indicated stable binding affinity between CDC42 and the contaminants (Fig. 13). 4. Discussion The pervasive presence of contaminants within the global food supply has sparked considerable apprehension regarding their detrimental effects on human health, particularly their propensity to trigger carcinogenesis 35 . Although extensive research has been conducted on the toxicological profiles of food contaminants, such as Glyphosate, PFOS, Nitrosamines, and others 36 – 38 , the molecular pathways through which these agents promote the development of cancer remain inadequately elucidated. In this study, we harnessed a comprehensive array of advanced databases and bioinformatics network methodologies, including PubChem, ADEMTlab3.0, Protox3, GeneCards, OMIM, STRING, Machine Learning, and WGCNA, to identify and analyze relevant targets. Following this, we utilized Cytoscape software for an in-depth analysis of these targets. The DAVID database was then employed to perform GO and KEGG enrichment analyses on the core targets. Subsequently, we leveraged the PubChem database, in conjunction with AutoDock Vina, to conduct molecular docking studies,which facilitated an exploration of the intermolecular interactions between food contaminants and key target proteins. Collectively, our findings shed light on the potential molecular mechanisms by which food contaminants may contribute to the onset of carcinogenesis across various cancer types. Since 1970, a wide range of environmental toxins, such as chemicals in tobacco smoke, air pollutants, and contaminants in food have been identified 39 . These toxins are prevalent in daily life due to industrial growth and can impact the respiratory and cardiovascular systems and have been shown to be closely linked with the onset of various chronic diseases and cancers 40 . Dietary exposure to these toxins poses significant health risks. For this study, we selected nine common dietary contaminants (Glyphosate, PFOS, Nitrosamines, PBDEs, Methylmercury, Dioxins, Acrylamide, Pyrrolizidine Alkaloids, and Aflatoxin) based on data from the National Institute of Environmental Health Sciences (NIEHS) and the Environmental Protection Agency (EPA). Our comprehensive analysis identified 69 targets intersecting with pan-cancer pathways, which were further refined through PPI analysis and GO/KEGG enrichment analyses. We focused specifically on BRCA, PRAD, and COAD cancers to elucidate potential mechanisms linking these contaminants to carcinogenesis. In the context of BRCA, our study identified 49 hub targets associated with contaminant-induced BRCA. The GO and KEGG enrichment analyses for these hub targets revealed a multitude of cancer-related functions and pathways, including the PI3K-Akt pathway, which is well-documented in the development and progression of breast cancer 41 . This finding is consistent with previous studies, such as the work by Kavarthapu R et al., which demonstrated that JUN, a key target identified in our study, acts as a critical regulator of the PI3K-Akt pathway and plays a significant role in breast cancer development 42 . Similarly, EGFR, another key target identified in our study by machine learning algorithms and WGCNA, has been shown to drive tumor growth and metastasis in BRCA 43 . Our RNA sequencing analysis, based on TCGA gene expression data, further supports these findings by revealing significantly lower expression of JUN and EGFR in tumor samples compared to normal tissues. Additionally, higher JUN expression was correlated with longer survival in BRCA patients, although no significant differences were observed for EGFR expression levels. Molecular docking analysis demonstrated stable binding affinities between the contaminants and JUN, suggesting direct interactions that may contribute to carcinogenesis. These results not only corroborate prior research but also provide novel insights into the molecular mechanisms underlying contaminant-induced BRCA, highlighting JUN as potential therapeutic targets. Concerning PRAD, our research discovered 44 hub targets related to food contaminant-induced prostate cancer. The GO and KEGG enrichment analyses emphasized the considerable involvement of pathways, such as the PI3K-Akt pathway, which is widely documented in the progression of prostate cancer 44 . This discovery is in line with recent studies that demonstrate the role of PI3K-Akt signaling in prostate cancer, particularly in regulating cell proliferation and apoptosis 45 . Specifically, our analysis identified MAPK14 and CDC42 as crucial targets, which have been associated with prostate cancer through their functions in inflammatory responses and DNA repair mechanisms. RNA sequencing analysis based on the TCGA data revealed a significantly decreased expression of MAPK14 in tumor samples compared to normal tissues, with lower MAPK14 expression being correlated with longer survival in PRAD patients. However, no significant disparities were observed in CDC42 expression levels between tumor and normal tissues from the same patient in the TCGA data, nor in the survival analysis. This led us to further validate MAPK14 as a potential therapeutic target. Molecular docking analysis demonstrated stable binding affinities between food contaminants and MAPK14, suggesting direct interactions that might contribute to carcinogenesis. This finding extends previous research by providing evidence of a molecular connection between environmental contaminants and key oncogenic pathways in prostate cancer. Furthermore, 48 core targets were recognized to be associated with food contaminant-induced COAD, disclosing the significant involvement of cancer-related pathways such as transcriptional regulation and apoptosis. These discoveries are consistent with previous studies that have emphasized the crucial role of these pathways in the development of COAD 46 , 47 . Nevertheless, our study uniquely combines food contaminant targets with COAD-related genes, offering novel insights into potential environmental triggers. For example, our analysis identified CDC42 as a key target with significantly lower expression in tumor samples compared to normal tissues, and lower CDC42 expression was associated with worse survival outcomes. This is in line with recent studies indicating that the dysregulation of CDC42 is connected to aggressive cancer phenotypes 48 . In contrast, the expression of ESR2 did not demonstrate a significant correlation with survival, suggesting a more intricate role that requires further examination. Molecular docking analysis further disclosed stable binding affinities between food contaminants and CDC42, suggesting direct interactions that might contribute to carcinogenesis. These results not only support previous research but also introduce new potential therapeutic targets and biomarkers for COAD, emphasizing the significance of considering environmental factors in cancer development. Our study innovatively integrates PPI analysis, machine learning algorithms, WGCNA, and molecular docking simulations to provide a comprehensive understanding of the molecular mechanisms underlying contaminant-induced carcinogenesis. This integrative approach not only identifies key targets and pathways significantly associated with food contaminants and cancer development but also reveals the intricate interactions between environmental factors and cellular processes. For instance, the identification of critical regulatory genes such as JUN and CDC42, which exhibit altered expression in tumor tissues and correlate with survival outcomes, offers novel insights for targeted therapy development. These findings could facilitate the design of small molecules or biologics that specifically inhibit the activity of these targets, thereby mitigating the carcinogenic effects of food contaminants.Moreover, our study highlights the potential for public health interventions. By pinpointing specific contaminants that interact with key cancer-related genes, our research provides a basis for developing policies aimed at reducing exposure to harmful substances in the food supply. This could significantly lower the incidence of contaminant-induced cancers and other related diseases, thereby enhancing overall public health. Additionally, the identified targets and pathways may serve as biomarkers for early detection and monitoring of cancer, further supporting personalized medicine approaches. Our study thus bridges the gap between environmental exposures and molecular mechanisms, offering both therapeutic and preventive strategies that could substantially impact cancer management and public health initiatives. Despite the comprehensive nature of our study, several limitations should be acknowledged. First, our analysis relies heavily on in silico methods, and the identified targets and pathways need to be validated through experimental studies. Future work should focus on validating the molecular interactions identified through molecular docking simulations using in vitro and in vivo models. Additionally, the potential synergistic effects of multiple contaminants, which are often present in real-world scenarios, were not fully explored in this study. Future research should investigate the combined effects of contaminants on carcinogenesis, as this could provide more accurate insights into the risks posed by contaminated food. 5. Conclusion In conclusion, our study provides novel insights into the molecular mechanisms linking food contaminants to carcinogenesis, identifying key targets and pathways that are significantly associated with contaminant-induced cancers. The findings highlight the potential of JUN, CDC42, and MAPK14 as therapeutic targets for mitigating the carcinogenic effects of food contaminants. While our study has several limitations, it lays the groundwork for future research aimed at validating these findings and exploring their potential applications in cancer prevention and treatment. Ultimately, a better understanding of the molecular underpinnings of contaminant-induced carcinogenesis could lead to more effective strategies for reducing the burden of cancer worldwide. Declarations CRediT authorship contribution statement Bangsheng Chen: Visualization, Conceptualization. Maomao Li: Software, Methodology. Wenzhu Lou: Investigation. Zhiyan Wang: Methodology, Investigation. Yi Gu: Formal analysis. Feiyan Mao and Lian Tan: Writing - review & editing, Data curation. Shuaishuai Huang: Writing - original draft, Visualization, Methodology, Conceptualization. Funding This study was supported by the Natural Science Foundation of Ningbo, China (2022J039 to C.B.). Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Declaration of generative AI and AI-assisted technologies in the writing process. During the preparation of this work, we used deepseek in order to improve the readability and language of the manuscript. After using this tool, we reviewed and edited the content as needed and take full responsibility for the content of the published article. Author Contribution All authors reviewed the manuscript References F. Bray, et al., Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries, CA Cancer J Clin 74(3) 229–263 (2024). K. Inamura, et al., Cancer as microenvironmental, systemic and environmental diseases: opportunity for transdisciplinary microbiomics science, Gut (2022). V. Kayamba, P. Kelly, Environmental factors associated with gastric carcinogenesis, Curr Opin Gastroenterol 38(2) 156–161 (2022). A. Maher, A. Nowak, Chemical Contamination in Bread from Food Processing and Its Environmental Origin, Molecules 27(17) (2022). P. Sadighara, et al., Residues of carcinogenic pesticides in food: a systematic review, Rev Environ Health 39(4) 659–666 (2024). M. Masci, et al., Chromatographic Methods for the Determination of Glyphosate in Cereals Together with a Discussion of Its Occurrence, Accumulation, Fate, Degradation, and Regulatory Status, Methods Protoc 7(3) (2024). T. Bonato, et al., Contamination of the terrestrial food chain by per- and polyfluoroalkyl substances (PFAS) and related human health risks: A systematic review, Sci Total Environ 961 178337 (2025). A. Vignesh, et al., Food contaminants: Impact of food processing, challenges and mitigation strategies for food security, Food Res Int 191 114739 (2024). P. Sipos, et al., Physical and Chemical Methods for Reduction in Aflatoxin Content of Feed and Food, Toxins (Basel) 13(3) (2021). L. Zeng, et al., Mitochondrial Dysfunction-Molecular Mechanisms and Potential Treatment approaches of Hepatocellular Carcinoma, Mol Cell Biochem 480(4) 2131–2142 (2025). M. Sassano, et al., Exposure to per- and poly-fluoroalkyl substances and lung, head and neck, and thyroid cancer: A systematic review and meta-analysis, Environ Res 266 120606 (2025). L. Bridgeman, et al., Effect of Acrylamide and Mycotoxins in SH-SY5Y Cells: A Review, Toxins (Basel) 16(2) (2024). M. Cheng, et al., Exploring the mechanism of PPCPs on human metabolic diseases based on network toxicology and molecular docking, Environ Int 196 109324 (2025). Y. Li, et al., Air pollution and prostate cancer: Unraveling the connection through network toxicology and machine learning, Ecotoxicol Environ Saf 292 117966 (2025). N. He, et al., Elucidating the mechanism of plasticizers inducing breast cancer through network toxicology and molecular docking analysis, Ecotoxicol Environ Saf 284 116866 (2024). Z. Xu, et al., Identification of molecular targets of Hypericum perforatum in blood for major depressive disorder: a machine-learning pharmacological study, Chin Med 19(1) 141 (2024). P.C. Agu, et al., Molecular docking as a tool for the discovery of molecular targets of nutraceuticals in diseases management, Sci Rep 13(1) 13398 (2023). M.A. Abbas, et al., Comprehensive understanding of the role of GPER in estrogen receptor-alpha negative breast cancer, J Steroid Biochem Mol Biol 241 106523 (2024). B. Arjmand, et al., Molecular Docking as a Therapeutic Approach for Targeting Cancer Stem Cell Metabolic Processes, Front Pharmacol 13 768556 (2022). S. Kim, et al., PubChem 2023 update, Nucleic Acids Research 51(D1) D1373-D1380%@ 0305–1048 (2022). L. Fu, et al., ADMETlab 3.0: an updated comprehensive online ADMET prediction platform enhanced with broader coverage, improved performance, API functionality and decision support, Nucleic Acids Res 52(W1) W422-W431 (2024). P. Banerjee, et al., ProTox 3.0: a webserver for the prediction of toxicity of chemicals, Nucleic Acids Res 52(W1) W513-W520 (2024). M. Kuhn, et al., STITCH: interaction networks of chemicals and proteins, Nucleic Acids Res 36(Database issue) D684-8 (2008). D. Gfeller, et al., SwissTargetPrediction: a web server for target prediction of bioactive small molecules, Nucleic Acids Res 42(Web Server issue) W32-8 (2014). C. Chen, et al., SEA version 3.0: a comprehensive extension and update of the Super-Enhancer archive, Nucleic Acids Res 48(D1) D198-D203 (2020). J.S. Amberger, et al., OMIM.org: Online Mendelian Inheritance in Man (OMIM(R)), an online catalog of human genes and genetic disorders, Nucleic Acids Res 43(Database issue) D789-98 (2015). G. Stelzer, et al., The GeneCards Suite: From Gene Data Mining to Disease Genome Sequence Analyses, Curr Protoc Bioinformatics 54 1 30 1–1 30 33 (2016). D. Szklarczyk, et al., The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets, Nucleic Acids Res 49(D1) D605-D612 (2021). P. Shannon, et al., Cytoscape: a software environment for integrated models of biomolecular interaction networks, Genome Res 13(11) 2498–504 (2003). B.T. Sherman, et al., DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update), Nucleic Acids Res 50(W1) W216-W221 (2022). K. Yang, et al., Machine learning-optimized targeted detection of alternative splicing, Nucleic Acids Res 53(3) (2025). P. Langfelder, S. Horvath, WGCNA: an R package for weighted correlation network analysis, BMC Bioinformatics 9 559 (2008). C. UniProt, UniProt: the Universal Protein Knowledgebase in 2023, Nucleic Acids Res 51(D1) D523-D531 (2023). M. Bugnon, et al., SwissDock 2024: major enhancements for small-molecule docking with Attracting Cavities and AutoDock Vina, Nucleic Acids Res 52(W1) W324-W332 (2024). B.R. Rushing, M.I. Selim, Aflatoxin B1: A review on metabolism, toxicity, occurrence in food, occupational exposure, and detoxification methods, Food Chem Toxicol 124 81–100 (2019). D. Soares, et al., Glyphosate Use, Toxicity and Occurrence in Food, Foods 10(11) (2021). B. Zhang, et al., Impacts of PFOS, PFOA and their alternatives on the gut, intestinal barriers and gut-organ axis, Chemosphere 361 142461 (2024). M. Alikord, et al., Food safety and quality assessment: comprehensive review and recent trends in the applications of ion mobility spectrometry (IMS), Crit Rev Food Sci Nutr 62(18) 4833–4866 (2022). L.B. Murrison, et al., Environmental exposures and mechanisms in allergy and asthma development, J Clin Invest 129(4) 1504–1515 (2019). H. Orru, et al., The Interplay of Climate Change and Air Pollution on Health, Curr Environ Health Rep 4(4) 504–513 (2017). I.M. Browne, et al., Optimal targeting of PI3K-AKT and mTOR in advanced oestrogen receptor-positive breast cancer, Lancet Oncol 25(4) e139-e151 (2024). R. Kavarthapu, et al., Crosstalk between PRLR and EGFR/HER2 Signaling Pathways in Breast Cancer, Cancers (Basel) 13(18) (2021). S. Sigismund, et al., Emerging functions of the EGFR in cancer, Mol Oncol 12(1) 3–20 (2018). B.Y. Shorning, et al., The PI3K-AKT-mTOR Pathway and Prostate Cancer: At the Crossroads of AR, MAPK, and WNT Signaling, Int J Mol Sci 21(12) (2020). M. Hashemi, et al., Targeting PI3K/Akt signaling in prostate cancer therapy, J Cell Commun Signal 17(3) 423–443 (2023). J. Bian, et al., Transcriptional Regulation of Wnt/beta-Catenin Pathway in Colorectal Cancer, Cells 9(9) (2020). P. Zhou, et al., The Induction Mechanism of Ferroptosis, Necroptosis, and Pyroptosis in Inflammatory Bowel Disease, Colorectal Cancer, and Intestinal Injury, Biomolecules 13(5) (2023). V. Justilien, et al., Oncogenic Ect2 signaling regulates rRNA synthesis in NSCLC, Small GTPases 10(5) 388–394 (2019). Additional Declarations No competing interests reported. Supplementary Files SupplementaryFigures.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6625099","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":461106234,"identity":"7c56c8c9-74b6-4ef3-95c1-643a5cd0c091","order_by":0,"name":"Bangsheng Chen","email":"","orcid":"","institution":"Ningbo Yinzhou No. 2 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Bangsheng","middleName":"","lastName":"Chen","suffix":""},{"id":461106237,"identity":"731c54ad-59a2-4334-bc1f-80626d95b537","order_by":1,"name":"Maomao Li","email":"","orcid":"","institution":"Ningbo Yinzhou No. 2 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Maomao","middleName":"","lastName":"Li","suffix":""},{"id":461106238,"identity":"f3526731-41ce-4175-a567-33b74d74a9c0","order_by":2,"name":"Yi Gu","email":"","orcid":"","institution":"The Affiliated Lihuili Hospital of Ningbo University","correspondingAuthor":false,"prefix":"","firstName":"Yi","middleName":"","lastName":"Gu","suffix":""},{"id":461106239,"identity":"19f60d14-7d95-415f-900f-dc98e2d6c866","order_by":3,"name":"Wenzhu Lou","email":"","orcid":"","institution":"Ningbo Yinzhou No. 2 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Wenzhu","middleName":"","lastName":"Lou","suffix":""},{"id":461106240,"identity":"60c153cb-a610-4f1a-a514-90a978ef98f1","order_by":4,"name":"Shuaishuai Huang","email":"","orcid":"","institution":"Ningbo Yinzhou No. 2 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Shuaishuai","middleName":"","lastName":"Huang","suffix":""},{"id":461106243,"identity":"2ba0d989-a60f-4d2c-b14d-2db0a3e9cc14","order_by":5,"name":"Feiyan Mao","email":"","orcid":"","institution":"Ningbo No. 2 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Feiyan","middleName":"","lastName":"Mao","suffix":""},{"id":461106244,"identity":"2d8939b0-30a1-4e0e-ae31-ae227fdbaf85","order_by":6,"name":"Lian Tan","email":"","orcid":"","institution":"Ningbo Yinzhou No. 2 Hospital","correspondingAuthor":false,"prefix":"","firstName":"Lian","middleName":"","lastName":"Tan","suffix":""},{"id":461106248,"identity":"b2047c99-b8ca-444a-87aa-065ba958c93c","order_by":7,"name":"Zhiyan Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIie3PsQqCQBzH8f9xkMtpa071CP8QIkh6FuXAlgihxVE4qKUH8DF6g7Q/2SK0OjQ3Cy0NDaVjgzo23Jebjt+H4wB0un+MNwe8IWdxWqG77E/svcqyJAxk34fAwyKXJKozi7vmeDXpGUa0gXKN5GLKwaDLsZWQ5TlJQVuWfMka7xaIICjbiUBp7shXo4Y8OIzErJNQTXY1mSOxuAeZqpocRO4R9CE2CYeLYuUnhkqzAwZy0PUX61Y4TxEt/BMxVb3e7nJoUN5KJunvzaBtXjeOuxY6nU6n+wCedVCIuKr8VwAAAABJRU5ErkJggg==","orcid":"","institution":"Ningbo Yinzhou No. 2 Hospital","correspondingAuthor":true,"prefix":"","firstName":"Zhiyan","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2025-05-09 05:08:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6625099/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6625099/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":83553573,"identity":"8ac0f45f-72bf-4765-ab86-15f140e52eb8","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":292769,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eThe associations between food contaminants and pan-cancer.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Venn Diagram depicting the overlap of targets associated with food contaminants and pan-cancer.\u003c/p\u003e\n\u003cp\u003e(B) Visualization of interacting targets using Cytoscape 3.10.3, illustrating the network of target interactions.\u003c/p\u003e\n\u003cp\u003e(C) The potential hub genes identified by the Cytoscape 2.2 plugin are arranged in descending order of degree.\u003c/p\u003e\n\u003cp\u003e(D) GO enrichment analysis highlighting significant biological processes, cellular components, and molecular functions related to the interact targets.\u003c/p\u003e\n\u003cp\u003e(E, F) KEGG enrichment analysis of the interact targets, identifying enriched pathways.\u003c/p\u003e\n\u003cp\u003e(G) Compound-target-pathway network diagram of food contaminants induced pan cancer.\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/45ea0d6c59dbab6941ed498f.jpg"},{"id":83553577,"identity":"a566b371-6206-4c7e-9fcb-947fbf87a853","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":305299,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAnalysis of associations between food contaminants and breast cancer.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) The Venn diagram of related targets between food contaminants and breast cancer.\u003c/p\u003e\n\u003cp\u003e(B) PPI network diagram of core targets.\u003c/p\u003e\n\u003cp\u003e(C) MCODE classification of core targets.\u003c/p\u003e\n\u003cp\u003e(D) The top 20 KEGG enrichment results displayed in gene ratio order and the relationship between the top ten hub genes and pathways.\u003c/p\u003e\n\u003cp\u003e(E) Top 10 Hubba gene of core targets.\u003c/p\u003e\n\u003cp\u003e(F) GO enrichment results.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/13cacfbaebc791a366a3b085.jpg"},{"id":83553575,"identity":"f422f496-7f08-4a9c-a211-5849d3e2590c","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":193708,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMachine learning-based screening of key targets.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) LASSO coefficient profiles. The LASSO model was applied for regression of high-dimensional predictors. The method uses an L1 penalty to shrink some regression coefficients to exactly zero, thereby selecting the most significant predictors. Each curve represents the trajectory of a predictor’s coefficient as λ varies.\u003c/p\u003e\n\u003cp\u003e(B) Lasso coefficient distribution misclassification error.\u003c/p\u003e\n\u003cp\u003e(C) Feature importance of the random forest model\u003c/p\u003e\n\u003cp\u003e(D) Boruta-based feature selection was used to sort the importance of the features.\u003c/p\u003e\n\u003cp\u003e(E) SVM-RFE model.\u003c/p\u003e\n\u003cp\u003e(F) Venn diagram of four algorithms for screening key targets.\u003c/p\u003e\n\u003cp\u003e(G) The nomogram is based on the six-gene signature.\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/5d3e89e81781a9f77babb89f.jpg"},{"id":83553576,"identity":"3f8f98ac-b956-435f-b8b1-461c024fb8c9","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":301115,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCharacterization of hub targets for expression, survival probability, and gene set enrichment analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Expression of hub targets in tumors and normal tissues in the TCGA-BRCA dataset, and Wilcoxon rank sum was used for the significance test.\u003c/p\u003e\n\u003cp\u003e(B, D) Differentiated expression of JUN and EGFR in the tumor and normal sample. Analyses were performed across all normal and tumor samples with p value closing to zero by Wilcoxon rank sum test.\u003c/p\u003e\n\u003cp\u003e(C, E) Paired differentiation analysis for expression of JUN and EGFR in the normal and tumor sample deriving from the same one patient.\u003c/p\u003e\n\u003cp\u003e(F, G) Survival analysis for BRCA patients with different targets expression. Patients were labeled with higher quartile versus lower quartile groups defined by expression of each gene indicated.\u003c/p\u003e\n\u003cp\u003e(H, I) GSEA enrichment results for JUN.\u003c/p\u003e","description":"","filename":"Picture4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/3ac4b8812aa58104d8579912.jpg"},{"id":83554383,"identity":"1a3e217f-39d1-4ede-be39-9812099e3ae8","added_by":"auto","created_at":"2025-05-28 11:15:49","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":237911,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMolecular docking results of the lowest binding energy in 9 food comtaminants with the JUN.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/2a90b0fa3f1b84a98c5137e9.jpg"},{"id":83555069,"identity":"e41a8002-06f2-475d-8d6b-ac010353a055","added_by":"auto","created_at":"2025-05-28 11:23:49","extension":"jpg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":338949,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAnalysis of associations between food contaminants and prostate cancer.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) The Venn diagram of related targets between food contaminants and prostate cancer.\u003c/p\u003e\n\u003cp\u003e(B) PPI network diagram of core targets.\u003c/p\u003e\n\u003cp\u003e(C) MCODE classification of core targets.\u003c/p\u003e\n\u003cp\u003e(D) The top 20 KEGG enrichment results displayed in gene ratio order and the relationship between the top ten hub genes and pathways.\u003c/p\u003e\n\u003cp\u003e(E) Top 10 Hubba gene of core targets.\u003c/p\u003e\n\u003cp\u003e(F) GO enrichment results.\u003c/p\u003e","description":"","filename":"Picture6.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/a80ae54074ca6facd9c66799.jpg"},{"id":83554386,"identity":"4b186fa8-eda8-4029-9f84-fd737696f2f2","added_by":"auto","created_at":"2025-05-28 11:15:49","extension":"jpg","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":221209,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMachine learning-based screening of key targets.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) LASSO coefficient profiles. The LASSO model was applied for regression of high-dimensional predictors.\u003c/p\u003e\n\u003cp\u003e(B) Lasso coefficient distribution misclassification error.\u003c/p\u003e\n\u003cp\u003e(C) Feature importance of the random forest model\u003c/p\u003e\n\u003cp\u003e(D) Boruta-based feature selection was used to sort the importance of the features.\u003c/p\u003e\n\u003cp\u003e(E) SVM-RFE model.\u003c/p\u003e\n\u003cp\u003e(F) Venn diagram of four algorithms for screening key targets.\u003c/p\u003e\n\u003cp\u003e(G) The nomogram is based on the fifteen-gene signature.\u003c/p\u003e","description":"","filename":"Picture7.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/a093433808a31fc387b7611f.jpg"},{"id":83554388,"identity":"359044e3-67f0-44ef-bf7e-4fc356675285","added_by":"auto","created_at":"2025-05-28 11:15:49","extension":"jpg","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":299026,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCharacterization of hub targets for expression, survival probability, and gene set enrichment analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Expression of hub targets in tumors and normal tissues in the TCGA-PRAD dataset, and Wilcoxon rank sum was used for the significance test.\u003c/p\u003e\n\u003cp\u003e(B, D) Differentiated expression of MAPK14and CDC42 in the tumor and normal sample. Analyses were performed across all normal and tumor samples with p value closing to zero by Wilcoxon rank sum test.\u003c/p\u003e\n\u003cp\u003e(C, E) Paired differentiation analysis for expression of MAPK14and CDC42 in the normal and tumor sample deriving from the same one patient.\u003c/p\u003e\n\u003cp\u003e(F, G) Survival analysis for PRAD patients with different targets expression. Patients were labeled with higher quartile versus lower quartile groups defined by expression of each gene indicated.\u003c/p\u003e","description":"","filename":"Picture8.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/dbd220674fbea53e0de1e1f9.jpg"},{"id":83553586,"identity":"108afac6-4b49-4d4b-bcd0-c37623a36472","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":250978,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMolecular docking results of the lowest binding energy in 9 food comtaminants with the MAPK14.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture9.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/5bba77edc41f545717c3cc10.jpg"},{"id":83553581,"identity":"68193b12-6531-4488-a5e3-08a8c384fd71","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":323461,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAnalysis of associations between food contaminants and colorectal cancer.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) The Venn diagram of related targets between food contaminants and colorectal cancer.\u003c/p\u003e\n\u003cp\u003e(B) PPI network diagram of core targets.\u003c/p\u003e\n\u003cp\u003e(C) MCODE classification of core targets.\u003c/p\u003e\n\u003cp\u003e(D) The top 20 KEGG enrichment results displayed in gene ratio order and the relationship between the top ten hub genes and pathways.\u003c/p\u003e\n\u003cp\u003e(E) Top 10 Hubba gene of core targets.\u003c/p\u003e\n\u003cp\u003e(F) GO enrichment results.\u003c/p\u003e","description":"","filename":"Picture10.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/634b88331dcffaa8189ee56d.jpg"},{"id":83553582,"identity":"0fa9310f-41a8-41c7-b071-cf8c528cb74b","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":204334,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMachine learning-based screening of key targets.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) LASSO coefficient profiles. The LASSO model was applied for regression of high-dimensional predictors.\u003c/p\u003e\n\u003cp\u003e(B) Lasso coefficient distribution misclassification error.\u003c/p\u003e\n\u003cp\u003e(C) Feature importance of the random forest model\u003c/p\u003e\n\u003cp\u003e(D) Boruta-based feature selection was used to sort the importance of the features.\u003c/p\u003e\n\u003cp\u003e(E) SVM-RFE model.\u003c/p\u003e\n\u003cp\u003e(F) Venn diagram of four algorithms for screening key targets.\u003c/p\u003e\n\u003cp\u003e(G) The nomogram is based on the fifteen-gene signature.\u003c/p\u003e","description":"","filename":"Picture11.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/4065ecb1d094f7648a9be640.jpg"},{"id":83553588,"identity":"c280d8b0-a778-4ef7-89f2-b8bc2c1630c5","added_by":"auto","created_at":"2025-05-28 11:07:49","extension":"jpg","order_by":12,"title":"Figure 12","display":"","copyAsset":false,"role":"figure","size":295080,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCharacterization of hub targets for expression, survival probability, and gene set enrichment analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Expression of hub targets in tumors and normal tissues in the TCGA-PRAD dataset, and Wilcoxon rank sum was used for the significance test.\u003c/p\u003e\n\u003cp\u003e(B, D) Differentiated expression of MAPK14and CDC42 in the tumor and normal sample. Analyses were performed across all normal and tumor samples with p value closing to zero by Wilcoxon rank sum test.\u003c/p\u003e\n\u003cp\u003e(C, E) Paired differentiation analysis for expression of MAPK14and CDC42 in the normal and tumor sample deriving from the same one patient.\u003c/p\u003e\n\u003cp\u003e(F, G) Survival analysis for PRAD patients with different targets expression. Patients were labeled with higher quartile versus lower quartile groups defined by expression of each gene indicated.\u003c/p\u003e","description":"","filename":"Picture12.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/7485a4f8d62f91105734324d.jpg"},{"id":83554387,"identity":"fb6fb8b5-40b4-4108-a5c3-b40c5b63f8cc","added_by":"auto","created_at":"2025-05-28 11:15:49","extension":"jpg","order_by":13,"title":"Figure 13","display":"","copyAsset":false,"role":"figure","size":257770,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMolecular docking results of the lowest binding energy in 9 food comtaminants with the CDC42.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Picture13.jpg","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/7dd057e3a5bd011ef41fcc00.jpg"},{"id":87226589,"identity":"286a605a-ecb4-43ed-b01f-548321c13835","added_by":"auto","created_at":"2025-07-21 17:31:45","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":5072942,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/88d5eeaf-13ac-4dba-8527-38f45a750b4b.pdf"},{"id":83555070,"identity":"57e8bf8a-398e-4ee8-a844-6a9eba48cd88","added_by":"auto","created_at":"2025-05-28 11:23:49","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":686642,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigures.docx","url":"https://assets-eu.researchsquare.com/files/rs-6625099/v1/7c5704aeeb1b314f6bb7316e.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Unraveling the Carcinogenic Mechanisms of Food contaminants through Network Toxicology, Machine Learning, and Molecular Docking","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eCancer remains a leading cause of mortality worldwide, accounting for nearly 10\u0026nbsp;million deaths annually, with projections suggesting a 50% increase in incidence by 2050\u003csup\u003e1\u003c/sup\u003e. Although advancements in oncology have shed light on the genetic and molecular drivers of carcinogenesis, approximately 20\u0026ndash;30% of cancers are linked to modifiable environmental factors, including chemical exposures\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Despite increasing awareness, public understanding of pervasive environmental carcinogens - such as air pollutants, industrial chemicals, and food contaminants - remains limited. This gap in knowledge perpetuates preventable exposures, contributing to the rising incidence of cancer and highlighting the urgent need to identify and mitigate these hazards.\u003c/p\u003e \u003cp\u003eAmong environmental carcinogens, food contaminants constitute a significant yet often underestimated threat\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Compounds such as glyphosate (herbicide residues), perfluorooctane sulfonate (PFOS, a persistent organic pollutant), nitrosamines (byproducts of processed meats), and aflatoxins (mycotoxins found in improperly stored crops) are prevalent throughout global food supply chains\u003csup\u003e\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. Epidemiological research has established direct associations between these substances and various malignancies: for instance, aflatoxin B1 is implicated in the induction of hepatocellular carcinoma through mechanisms involving apoptosis and autophagy, among others\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. Furthermore, exposure to PFOS has been linked to an increased risk of thyroid and lung cancers\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. Acrylamide, which forms during high-temperature cooking processes, is classified by the International Agency for Research on Cancer (IARC) as a Group 2A carcinogen due to its genotoxic properties\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. Despite the substantial body of evidence supporting these associations, regulatory measures have not kept pace with scientific advancements, highlighting the urgent need for mechanistic insights to guide policy development and public health strategies.\u003c/p\u003e \u003cp\u003eTraditional toxicology frequently encounters limitations in deciphering the multi - target mechanisms of foodborne carcinogens. Network toxicology addresses this limitation by integrating systems biology and computational modeling to map chemical-gene-disease interactions, thereby identifying central hubs within carcinogenic pathways\u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. For instance, Li et al. (2025) employed PPI networks and machine learning to identify key genes (HDAC6, CDK1, DNMT1, NOS3, DPP4) that mediate the effects of air pollutants on prostate cancer\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Similarly, He et al. (2024) applied network toxicology, protein interactions, and molecular docking to elucidate the interactions between plasticizers and key proteins implicated in breast cancer\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. These studies effectively address the \"needle-in-a-haystack\" challenge of target prioritization. Additionally, Xu et al. (2024) utilized differential expression analysis, WGCNA, database mining, machine learning, and molecular docking to elucidate the therapeutic mechanisms of Hypericum perforatum in major depressive disorder (MDD)\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Their integrated approach identified key targets and pathways, highlighting immune-related effects and providing a robust framework to translate high-dimensional data into actionable insights for disease prevention. These methodologies collectively bridge the gap between chemical exposure and phenotypic outcomes.\u003c/p\u003e \u003cp\u003eMolecular docking simulations offer atomic level insights into carcinogen-target interactions, complementing network based predictions\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. For instance, the study by Manal A. Abbas et al. (2024) demonstrated a high-affinity binding between bisphenol A and estrogen receptor α, shedding light on its role in the progression of breast cancer\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. Similarly, Babak Arjmand et al. (2019) utilized molecular docking to align ligand poses, rank active compounds, and propose potential anticancer agents\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. By integrating docking results with network topology, we can validate hub targets and refine mechanistic models, thus providing a robust toolkit for risk assessment.\u003c/p\u003e \u003cp\u003eThis study introduces a pioneering multi-modal framework that combines network toxicology, machine learning, WGCNA, and molecular docking to unravel foodborne carcinogenesis. We systematically examined nine high-risk contaminants, identified pan-cancer targets and pathway modules, and validated hub genes (JUN, CDC42, MAPK14) in breast, colon, and prostate cancers using TCGA cohorts. Our approach not only delineates novel mechanisms but also provides actionable targets for therapeutic development. By linking molecular signatures to clinical outcomes, this work underscores the imperative of food safety reforms and precision public health strategies to mitigate the global cancer burden.\u003c/p\u003e"},{"header":"2. Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1. Identification of physicochemical properties and toxicity of food contaminatants.\u003c/h2\u003e \u003cp\u003eWe obtained the chemical structures and corresponding molecular details of nine food contaminants from the PubChem database\u003csup\u003e20\u003c/sup\u003e(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubchem.ncbi.nlm.nih.gov\u003c/span\u003e\u003cspan address=\"https://pubchem.ncbi.nlm.nih.gov\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The carcinogenic properties of these pollutants were subsequently assessed using the ADMETLAB 3.0 platform\u003csup\u003e21\u003c/sup\u003e(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://admetlab3.scbdd.com\u003c/span\u003e\u003cspan address=\"https://admetlab3.scbdd.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and the ProTox3 database\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://tox.charite.de/protox3\u003c/span\u003e\u003cspan address=\"https://tox.charite.de/protox3\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) as evidenced in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2. Collection of food contaminants target genes\u003c/h2\u003e \u003cp\u003eThe potential human target genes associated with the nine food contaminants were sourced from the STITCH database\u003csup\u003e23\u003c/sup\u003e(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://stitch.embl.de/cgi/\u003c/span\u003e\u003cspan address=\"http://stitch.embl.de/cgi/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), the Swiss Target Prediction database\u003csup\u003e24\u003c/sup\u003e(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.swisstargetprediction.ch/\u003c/span\u003e\u003cspan address=\"http://www.swisstargetprediction.ch/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), the SEA database\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://sea.bkslab.org/\u003c/span\u003e\u003cspan address=\"https://sea.bkslab.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), and the TargetNet database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://targetnet.scbdd.com\u003c/span\u003e\u003cspan address=\"http://targetnet.scbdd.com\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The gene data obtainedfrom these four databases were consolidated, duplicates were eliminated, and a definitive set of target genes for the food contaminants was established.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3. Collection of cancer-related genes\u003c/h2\u003e \u003cp\u003eCancer-related genes were sourced from two databases: OMIM\u003csup\u003e26\u003c/sup\u003e(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://omim.org/\u003c/span\u003e\u003cspan address=\"https://omim.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and GeneCards\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.genecards.org/\u003c/span\u003e\u003cspan address=\"https://www.genecards.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). The genes were selected based on specific criteria: a \"score\" threshold equivalent to the median value from GeneCards was applied, and only genes exceeding this median score were chosen. Following the consolidation of gene lists from both databases and the subsequent removal of duplicate entries, a comprehensive set of cancer-associated genes was established.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e2.4. The intersection of food contaminants targets and cancerrelated targets\u003c/h2\u003e \u003cp\u003eThe intersection analysis between contaminant targets and cancer-related targets was conducted using the Venn diagram tool available on the web (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.bioinformatics.com.cn/\u003c/span\u003e\u003cspan address=\"http://www.bioinformatics.com.cn/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Both the contaminant target library and the cancer target library were uploaded to this platform, and the results were retrieved upon completion of the analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.5. Construction of protein - protein interaction (PPI) networks\u003c/h2\u003e \u003cp\u003eWe constructed the PPI network utilizing the STRING database\u003csup\u003e28\u003c/sup\u003e(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cn.string-db.org/)wit\u003c/span\u003e\u003cspan address=\"https://cn.string-db.org/)wit\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003eh overlapping targets identified from the Venn diagram. The\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eThe physicochemical properties and toxicity assessment of compounds\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"11\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCAS\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCompound Term\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMolecular Formula\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMolecular Weight(g/mol)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eLogP\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eLC\u003csub\u003e50\u003c/sub\u003eDM\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eLC\u003csub\u003e50\u003c/sub\u003eFM\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eIGC\u003csub\u003e50\u003c/sub\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eBCF\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eCarcinogenicity\u003c/p\u003e \u003cp\u003eIn ProTox 3.0\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eCarcinogenicity\u003c/p\u003e \u003cp\u003eIn ADMETLab3\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1071-83-6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGlyphosate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eC\u003csub\u003e3\u003c/sub\u003eH\u003csub\u003e8\u003c/sub\u003eNO\u003csub\u003e5\u003c/sub\u003eP\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e169.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-3.419\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e3.259\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e3.032\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e2.439\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.58\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.896\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1763-23-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePerfluorooctane sulfonate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eC\u003csub\u003e8\u003c/sub\u003eHF\u003csub\u003e17\u003c/sub\u003eO\u003csub\u003e3\u003c/sub\u003eS\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e500.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.817\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e6.341\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e5.744\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4.039\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e3.178\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.448\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e81407-93-4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNitrosamines\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eH\u003csub\u003e2\u003c/sub\u003eN\u003csub\u003e2\u003c/sub\u003eO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e46.029\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-0.617\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.769\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1.826\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e1.603\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.998\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1163-19-5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePentabromodiphenyl ethers\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eC\u003csub\u003e12\u003c/sub\u003eBr\u003csub\u003e10\u003c/sub\u003eO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e959.2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e6.977\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e7.666\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e9.13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e6.161\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.719\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.905\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e16056-34-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMethyl mercury\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCH\u003csub\u003e3\u003c/sub\u003eHg\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e215.63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.166\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e4.269\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e3.858\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e3.169\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e1.036\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.984\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1746-01-6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDioxins\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eC\u003csub\u003e12\u003c/sub\u003eH\u003csub\u003e4\u003c/sub\u003eCl\u003csub\u003e4\u003c/sub\u003eO\u003csub\u003e2\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e322\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e6.673\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e5.861\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e6.305\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e5.341\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e4.383\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.73\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e79-06-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAcrylamide\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eC\u003csub\u003e3\u003c/sub\u003eH\u003csub\u003e5\u003c/sub\u003eNO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e71.08\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e-0.513\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.779\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e2.861\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e2.307\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.522\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.714\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e81340-07-0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePyrrolizidine alkaloid\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eC\u003csub\u003e18\u003c/sub\u003eH\u003csub\u003e23\u003c/sub\u003eNO\u003csub\u003e5\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e333.4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e4.953\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e4.054\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e3.318\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.58\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1165-39-5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAflatoxin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eC\u003csub\u003e17\u003c/sub\u003eH\u003csub\u003e12\u003c/sub\u003eO\u003csub\u003e7\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e328.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.654\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e6.146\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e4.973\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e3.809\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e1.296\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.992\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e\u0026ldquo;CAS\u0026rdquo;: Chemical Abstracts Service; \u0026ldquo;LogP\u0026rdquo;: The logarithm of the n-octanol/water distribution coefficients at pH\u0026thinsp;=\u0026thinsp;7.4; \u0026ldquo;BCF\u0026rdquo;:Bioconcentration factors; \u0026ldquo;LC\u003csub\u003e50\u003c/sub\u003eDM\u0026rdquo;: 50% lethal concentrationin the daphnia magna after 48 h; \u0026ldquo;LC\u003csub\u003e50\u003c/sub\u003eFM\u0026rdquo;: 50% lethal concentration in the fathead minnow after 96 h; \u0026ldquo;IGC\u003csub\u003e50\u003c/sub\u003e\u0026rdquo;: 50% growth inhibition concentration in the tetrahymena pyriformis.\u003c/p\u003e \u003cp\u003especies was specified as Homo sapiens, and an interaction score of \u0026ge;\u0026thinsp;0.9 was applied to generate the PPI network diagram. The data obtained from STRING were imported into Cytoscape 3.10.3\u003csup\u003e29\u003c/sup\u003e for network visualization and analysis, facilitating the computation of topological properties and the creation of a PPI network diagram. Core targets were selected based on nodes exceeding the median values for betweenness centrality, closeness centrality, and degree. The MCODE plugin was utilized to identify the most significant subclusters of interacting nodes, and the CytoHubba plugin was employed to predict the top 10 significant genes using the maximal clique centrality (MCC) algorithm. Hub genes were ascertained by intersecting the targets of the MCODE significant module with those predicted by CytoHubba.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.6. GO and KEGG pathway analysis\u003c/h2\u003e \u003cp\u003eGO terms, encompassing biological processes, molecular functions, and cellular components, along with KEGG pathway analyses, were conducted utilizing the DAVID database\u003csup\u003e30\u003c/sup\u003e(\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://david.ncifcrf.gov/\u003c/span\u003e\u003cspan address=\"https://david.ncifcrf.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) to explore the pathways associated with cancer-related interactome targets. This analysis was designed to reveal and underscore the significant signaling pathways engaged in these biological processes. Visual analysis was implemented to efficiently interpret and display the outcomes of the GO and KEGG analyses, facilitated by the visualization platform provided at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.bioinformatics.com.cn\u003c/span\u003e\u003cspan address=\"https://www.bioinformatics.com.cn\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.7. Screening of hub targets using machine learning algorithms\u003c/h2\u003e \u003cp\u003eWe employed four distinct machine learning algorithms for target screening\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. Lasso regression analysis, SVM-R feature extraction (SVM-RFE), random forest (RandomForest), and Boruta feature selection were applied to the genes within the key subnetwork, using a random seed of \u0026ldquo;123\u0026rdquo; in RStudio (version 4.4.2). Genes identified at the intersection of these analyses were deemed as potential targets for toxicities associated with cancers. Furthermore, a nomogram depicting the five-year survival probability of cancer patients was constructed using the Rms package, demonstrating the clinical utility of these interactome targets.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e2.8. Identification of potential target genes through Weighted Gene Co-expression Network Analysis (WGCNA)\u003c/h2\u003e \u003cp\u003eFor the WGCNA analysis\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e, we sourced cancerous and normal tissues from the UCSC Xena database. We confirmed the soft threshold using the pickSoftThreshold function from the WGCNA package in R software. After establishing a scale-free network based on this soft threshold, a topology matrix and hierarchical clustering were applied. Subsequently, Eigengenes were identified for each module. Based on these Eigengenes, hierarchical clustering was performed once the correlation between modules was established. The module eigengene (ME), considered representative of the gene expression profiles, was calculated to identify modules associated with clinical traits. To pinpoint the most tumor-related modules, we conducted Module-Trait Relationships calculations for each module. Eigengene Adjacency Heatmaps were generated to illustrate module relationships.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e2.9. Characterization of key targets for expression, survival probability, and gene set enrichment analysis (GSEA)\u003c/h2\u003e \u003cp\u003eThe expression profiles of the identified key targets were correlated and visualized using the corrplot package within RStudio (version 4.4.3). The expression levels of each critical gene were assessed through the Wilcoxon rank sum test. Subsequently, GSEA was conducted for each key gene to gain further insights into the functions of the pathways enriched.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e2.10. Molecular docking\u003c/h2\u003e \u003cp\u003eWe integrated data from the Protein Data Bank (PDB) and UniProt databases\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e to acquire crystal structures of key targets, preprocessing them with AutoDock Vina version v1.2.7\u003csup\u003e34\u003c/sup\u003e. The PDB files utilized in this study were JUN (5FV8), MAPK14 (6SFO), and CDC42 (2QRZ). Detailed information on these protein structures is presented in Supplementary Table\u0026nbsp;1. The 3D structures of the small molecules were sourced from the PubChem database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubchem.ncbi.nlm.nih.gov/\u003c/span\u003e\u003cspan address=\"https://pubchem.ncbi.nlm.nih.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and stored in SDF format. The CAS numbers of the core toxic targets were employed to fetch their 3D structures from the PDB database (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.rcsb.org/\u003c/span\u003e\u003cspan address=\"http://www.rcsb.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), with Homo sapiens specified as the species and resolution refined between 0\u0026ndash;3.0 \u0026Aring;. Subsequently, we executed AutoDock for molecular docking of macromolecules and ligands using default parameters. Each docking pair was performed nine times, and the lowest binding affinity was documented. The docking models were visualized, and the binding energy was illustrated in a heat map using the pheatmap package within RStudio (version 4.4.3).\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec14\"\u003e\n \u003ch2\u003e3.1. Physicochemical and Toxicological Assessment of Contaminants\u003c/h2\u003e\n \u003cp\u003eA thorough evaluation of the physicochemical properties and toxicological profiles of nine representative contaminants was conducted utilizing systematic ADMET and ProTox analyses, as illustrated in Fig. S1. A significant positive correlation was observed between molecular weight and lipophilicity across the contaminants under study (Fig. S1A). Furthermore, the environmental toxicity of the contaminants was assessed through lethal concentrations (LC\u003csub\u003e50\u003c/sub\u003e) against Tetrahymena thermophila, Pimephales promelas, and Daphnia magna, revealing consistent variations in bioconcentration factors among the nine contaminants (Fig. S1B). Notably, the carcinogenic potential of these contaminants was found to be greater than 0.5 in both ProTox 3.0 and ADMETlab 3 platforms (Fig. S1C). These findings are consistent with existing clinical knowledge regarding the toxicity of the selected contaminants, confirming their particularly potent carcinogenic properties.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec15\"\u003e\n \u003ch2\u003e3.2. Identification of food contaminant-pan-cancer core targets and PPI network construction\u003c/h2\u003e\n \u003cp\u003eWe performed a comprehensive screening of 450 food contaminant targets from databases including STITCH, Swiss Target Prediction, SEA, and TargetNet, identifying 1852 pan-cancer targets from GeneCards and OMIM. A Venn diagram revealed 69 potential targets associated with contaminant-induced pan-cancer (Fig.\u0026nbsp;1A). Using STRING, we constructed PPI networks and refined them in Cytoscape with criteria like Closeness unDir\u0026thinsp;\u0026gt;\u0026thinsp;0.007993356128408048, Betweenness unDir\u0026thinsp;\u0026gt;\u0026thinsp;63.617647058823486, and Degree unDir\u0026thinsp;\u0026gt;\u0026thinsp;17.676470588235293, identifying 14 core targets (Figs.\u0026nbsp;1B, C).\u003c/p\u003e\n \u003cp\u003eWe then performed GO and KEGG enrichment analyses using DAVID, focusing on Homo sapiens. This generated 130 KEGG entries and 276 GO entries, including 167 biological processes, 22 cellular components, and 87 molecular functions. The top 10 GO entries with the highest gene ratios were highlighted in a bar chart (Fig. 1D), showing key processes like positive regulation of gene expression, chromatin remodeling, and positive regulation of transcription by RNA polymerase II. The top 20 KEGG entries were plotted (Figs. 1E,F), highlighting cancer-related pathways like pathways in cancer, PI3K-Akt signaling pathway, Proteoglycans in cancer.We selected three cancers for further analysis. Finally, we visualized the contaminant-target-pathway network in Cytoscape (Fig. 1G), with 9 contaminants (green), 20 pathways (orange), related target genes (blue), and 14 potential hub genes (red). This diagram clearly shows the complex relationships between contaminants, pathways, and target genes.\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec16\"\u003e\n \u003ch2\u003e3.3. The effect of food comtaminants on BRCA\u003c/h2\u003e\n \u003cp\u003eWe investigated the impact of food contaminants on breast cancer by integrating 15,632 targets from GeneCards and OMIM databases with 450 food contaminant targets, identifying 369 potential targets associated with food contaminant-induced breast cancer (Fig.\u0026nbsp;2A). Using the STRING database, we constructed PPI networks for these targets and imported them into Cytoscape for analysis. We applied criteria including Closeness unDir\u0026thinsp;\u0026gt;\u0026thinsp;0.0010247338113175926, Betweenness unDir\u0026thinsp;\u0026gt;\u0026thinsp;758.5291828793742, and Degree unDir\u0026thinsp;\u0026gt;\u0026thinsp;6.64591439688716 to select 49 core targets (Fig.\u0026nbsp;2B). The MCODE plugin in Cytoscape classified disease targets associated with breast cancer (Fig.\u0026nbsp;2C), and the CytoHubba plugin identified hub genes, revealing 5 subnetworks and the top 10 targets with the highest degree ranking (Fig.\u0026nbsp;2E).\u003c/p\u003e\n \u003cp\u003eGO and KEGG enrichment analysis using the DAVID database yielded 161 KEGG entries and 634 GO entries, including 441 biological processes (BP), 51 cellular components (CC), and 142 molecular functions (MF). The top 10 entries in BP, CC, and MF were selected for GO enrichment analysis, and the top 20 entries for KEGG analysis. The BP category was associated with transcription regulation by RNA polymerase II, positive gene expression regulation, apoptosis regulation, and chromatin remodeling. The CC category related to cytoplasm, cytosol, and nucleus, while MF concerned protein binding, enzyme binding, and identical protein binding (Fig.\u0026nbsp;2F). KEGG pathway analysis indicated significant involvement in cancer pathways, including pathways in cancer, proteoglycans in cancer, and chemical carcinogenesis-receptor activation (Fig.\u0026nbsp;2D). These insights shed light on the molecular mechanisms of food contaminant-induced breast cancer.\u003c/p\u003e\n \u003cp\u003eMachine learning algorithms were employed to identify key targets associated with breast cancer induced by food contaminants. Feature screening was executed by constructing LASSO regression models, identifying 31 genes (Figs. 3A, B). The SVM-RFE algorithm was also utilized to assess signature genes, resulting in the identification of 10 optimal signature genes (Fig. 3E). Random forest classification and Boruta feature importance algorithms were applied to select potential signature genes (Figs. 3C, D). By intersecting the signature genes obtained from these four models, six genes (PPARG, TLR4, EGFR, CASP3, CDK5, JUN) were pinpointed as key targets for subsequent breast cancer analysis (Fig. 3F). A nomogram depicting the five-year survival probability for breast cancer was crafted using the Rms package, underscoring the clinical utility of these targets (Fig. 3G).\u003c/p\u003e\n \u003cp\u003eWGCNA was performed using the TCGA-BRCA dataset to identify key gene modules. Samples were grouped based on Pearson correlation, with a soft threshold power set at 10 for optimal clustering (Figs. S2A, B). The dynamic tree cut package identified 17 distinct modules (Fig. S2C). Module-trait relationships were established to assess correlations with tumor occurrence. Significant positive correlations with tumor development were observed for the green-yellow (coefficient 0.46) and grey modules (coefficient 0.5), while the brown module showed a significant positive correlation with normal conditions (coefficient 0.66) (Fig. S2D). Eigengene adjacency heatmaps depicting these relationships are illustrated in Fig. S2E. This analysis provides potential targets into the molecular distinctions between tumor and normal tissue within BRCA. Interacting with hub genes selected by machine learning, two targets, JUN and EGFR, were identified.\u003c/p\u003e\n \u003cp\u003eRNA sequencing (RNAseq) data in the Transcripts Per Million (TPM) format from the UCSC Xena database were processed by the Toil process. Normal tissue data of BRCA and GTEX from TCGA were extracted, consisting of 179 normal and 1,099 tumor samples. RNAseq data were transformed into log2 format, and expression comparisons were performed (Fig. 4A). Wilcoxon rank sum test revealed that JUN and EGFR expression was significantly lower in tumor samples than in normal samples. Pairing analysis between normal and tumor tissues from the same patient showed similar results (Figs. 4B-E). Survival analysis indicated that BRCA patients with higher quartile JUN expression had longer survival than those with lower quartile JUN expression (Figs. 4F,G). GSEA functional analysis of JUN identified multiple associated pathways, including apical junction, epithelial-mesenchymal transition, hypoxia, e2f targets, and g2m checkpoint (Figs. 4H,I).\u003c/p\u003e\n \u003cp\u003eMolecular docking analysis was performed with JUN and 9 food contaminants. The lowest binding energies between these contaminants and the JUN protein were visualized. The contaminants interacted with key amino acid residues of JUN, including ARG-21, ARG-16, GLN-12, ARG-28, GLU-7, and GLU-15. A binding energy of -6.93 kcal/mol indicated stable binding affinity between JUN and the contaminants (Fig. 5).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec17\"\u003e\n \u003ch2\u003e3.4. \u003cem\u003eThe effect of food comtaminants on PRAD\u003c/em\u003e\u003c/h2\u003e\n \u003cp\u003eWe investigated the impact of food contaminants on prostate cancer by integrating 9,629 targets from GeneCards and OMIM databases with 450 food contaminant targets, identifying 323 potential targets associated with food contaminant-induced prostate cancer (Fig.\u0026nbsp;6A). Using the STRING database, we constructed protein-protein interaction (PPI) networks for these targets and imported them into Cytoscape for analysis. We applied criteria including Closeness unDir\u0026thinsp;\u0026gt;\u0026thinsp;0.0011772220504692277, Betweenness unDir\u0026thinsp;\u0026gt;\u0026thinsp;650.0086206896538, and Degree unDir\u0026thinsp;\u0026gt;\u0026thinsp;6.793103448275862 to select 44 core targets (Fig.\u0026nbsp;6B). The MCODE plugin in Cytoscape classified disease targets associated with prostate cancer (Fig.\u0026nbsp;6C), and the CytoHubba plugin identified hub genes, revealing 3 subnetworks and the top 10 targets with the highest degree ranking (Fig.\u0026nbsp;6E).\u003c/p\u003e\n \u003cp\u003eGO and KEGG enrichment analysis using the DAVID database yielded 163 KEGG entries and 606 GO entries, including 415 biological processes (BP), 53 cellular components (CC), and 138 molecular functions (MF). The top 10 entries in BP, CC, and MF were selected for GO enrichment analysis, and the top 20 entries for KEGG analysis. The BP category was associated with regulation of transcription by RNA polymerase II, regulation of gene expression, regulation of apoptotic process, and regulation of DNA-templated transcription. The CC category related to cytoplasm, nucleus, and cytosol, while MF concerned enzyme binding, identical protein binding, and protein kinase activity (Fig. 6F). KEGG pathway analysis indicated significant involvement in cancer pathways, including pathways in cancer, proteoglycans in cancer, and chemical carcinogenesis-receptor activation (Fig. 6D). These insights shed light on the molecular mechanisms of food contaminant-induced prostate cancer.\u003c/p\u003e\n \u003cp\u003eMachine learning algorithms were employed to identify key targets associated with prostatecancer induced by food contaminants. Feature screening was executed by constructing LASSO regression models, identifying 22 genes (Figs. 7A, B). The SVM-RFE algorithm was also utilized to assess signature genes, resulting in the identification of 30 optimal signature genes (Fig. 7E). Random forest classification and Boruta feature importance algorithms were applied to select potential signature genes (Figs. 7C, D). By intersecting the signature genes obtained from these four models, fifteen genes (APP, CDC42, CDK5, CYP19A1, EGFR, ESR1, ESR2, HDAC1, HSP90AA1, HSP90AB1, MAPK14, MMP9, PPARG, PRKACA, PTGS2) were pinpointed as key targets for subsequent prostate cancer analysis (Fig. 7F). A nomogram depicting the five-year survival probability for prostate cancer was crafted using the Rms package, underscoring the clinical utility of these targets (Fig. 7G).\u003c/p\u003e\n \u003cp\u003eWGCNA was performed using the TCGA-PRAD dataset to identify key gene modules. Samples were grouped based on Pearson correlation, with a soft threshold power set at 10 for optimal clustering (Figs. S3A, B). The dynamic tree cut package identified 14 distinct modules (Fig. S3C). Module-trait relationships were established to assess correlations with tumor occurrence. Significant positive correlations with tumor development were observed for the black (coefficient 0.32) and turquoise modules (coefficient 0.32), while the pink (coefficient 0.41) and magenta modules (coefficient 0.45) showed a significant positive correlation with normal conditions (Fig. S3D). Eigengene adjacency heatmaps depicting these relationships are illustrated in Fig. S3E. This analysis provides potential targets into the molecular distinctions between tumor and normal tissue within PRAD. Interacting with hub genes selected by machine learning, two targets, MAPK14,and CDC42were identified.\u003c/p\u003e\n \u003cp\u003eRNAseq data in the TPM format from the UCSC Xena database were processed using the Toil process. Data from normal tissues of PRAD and GTEX from The TCGA were extracted, encompassing 100 normal and 496 tumor samples. RNAseq data were normalized to the log2 scale, facilitating expression comparisons (Fig. 8A). The Wilcoxon rank sum test indicated significantly reduced expression of MAPK14 and CDC42 in tumor samples compared to normal samples. Pairwise analysis between matched normal and tumor tissues from the same patient corroborated these findings for MAPK14, though no significant difference was observed for CDC14 (Figs. 8B-E). Survival analysis revealed that patients with lower quartile MAPK14 expression in PRAD exhibited longer survival times than those with higher MAPK14 expression (Figs. 8F, G). GSEA of MAPK14 identified multiple associated pathways, including allograft rejection, epithelial-mesenchymal transition, inflammatory response, adipogenesis, DNA repair, and coagulation (Figs. 8H, I).\u003c/p\u003e\n \u003cp\u003eMolecular docking analysis was performed with MAPK14 and 9 food contaminants. The lowest binding energies between these contaminants and the MAPK14 protein were visualized. The contaminants interacted with key amino acid residues of MAPK14, including MET-109, HIS-107, SER-251, SER-252, LYS-249, TYR-132, ARG-136, ASN-82, LEU-151, ASP-150, THR-185, PRO-191, and LEU-195. A binding energy of -9.63 kcal/mol indicated stable binding affinity between MAPK14 and the contaminants (Fig. 9).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec18\"\u003e\n \u003ch2\u003e3.5. \u003cem\u003eThe effect of food comtaminants on COAD\u003c/em\u003e\u003c/h2\u003e\n \u003cp\u003eWe investigated the impact of food contaminants on colorectal cancer by integrating 11,550 targets from GeneCards and OMIM databases with 450 food contaminant targets, identifying 325 potential targets associated with food contaminant-induced colorectal cancer (Fig.\u0026nbsp;10A). Using the STRING database, we constructed PPI networks for these targets and imported them into Cytoscape for analysis. We applied criteria including Closeness unDir\u0026thinsp;\u0026gt;\u0026thinsp;0.0012390264023192084, Betweenness unDir\u0026thinsp;\u0026gt;\u0026thinsp;611.297777777777, and Degree unDir\u0026thinsp;\u0026gt;\u0026thinsp;6.844444444444444 to select 48 core targets (Fig.\u0026nbsp;10B). The MCODE plugin in Cytoscape classified disease targets associated with colorectal cancer (Fig.\u0026nbsp;10C), and the CytoHubba plugin identified hub genes, revealing 2 subnetworks and the top 10 targets with the highest degree ranking (Fig.\u0026nbsp;10E).\u003c/p\u003e\n \u003cp\u003eGO and KEGG enrichment analysis using the DAVID database yielded 162 KEGG entries and 630 GO entries, including 434 biological processes (BP), 55 cellular components (CC), and 141 molecular functions (MF). The top 10 entries in BP, CC, and MF were selected for GO enrichment analysis, and the top 20 entries for KEGG analysis. The BP category was associated with regulation of transcription by RNA polymerase II, regulation of gene expression, regulation of apoptotic process, and regulation of DNA-templated transcription. The CC category related to cytoplasm, nucleus, and cytosol, while MF concerned protein binding, identical protein binding, enzyme binding, and ATP binding (Fig. 10F). KEGG pathway analysis indicated significant involvement in cancer pathways, including pathways in cancer, proteoglycans in cancer, and chemical carcinogenesis - reactive oxygen species (Fig. 10D). These insights shed light on the molecular mechanisms of food contaminant-induced colorectal cancer.\u003c/p\u003e\n \u003cp\u003eMachine learning algorithms were employed to identify key targets associated with colorectal cancer induced by food contaminants. Feature screening was executed by constructing LASSO regression models, identifying 14 genes (Figs. 11A, B). The SVM-RFE algorithm was also utilized to assess signature genes, resulting in the identification of 30 optimal signature genes (Fig. 11E). Random forest classification and Boruta feature importance algorithms were applied to select potential signature genes (Figs. 11C, D). By intersecting the signature genes obtained from these four models, twelve genes (ABL1, AHR, CDC42, CYP19A1, ESR2, GSK3B, HSP90AB1, MMP9, NR3C1, PRKACA, PSEN1, RIPK2) were pinpointed as key targets for subsequent colorectal cancer analysis (Fig. 11F). A nomogram depicting the five-year survival probability for colorectal cancer was crafted using the Rms package, underscoring the clinical utility of these targets (Fig. 11G).\u003c/p\u003e\n \u003cp\u003eWGCNA was performed using the TCGA-COAD dataset to identify key gene modules. Samples were grouped based on Pearson correlation, with a soft threshold power set at 10 for optimal clustering (Figs. S4A, B). The dynamic tree cut package identified 21 distinct modules (Fig. S4C). Module-trait relationships were established to assess correlations with tumor occurrence. Significant positive correlations with tumor development were observed for the lightcyan (coefficient 0.57) and magenta modules (coefficient 0.38), while the yellow (coefficient 0.78) and turquoise modules (coefficient 0.72) showed a significant positive correlation with normal conditions (Fig. S4D). Eigengene adjacency heatmaps depicting these relationships are illustrated in Fig. S4E. This analysis provides potential targets into the molecular distinctions between tumor and normal tissue within COAD. Interacting with hub genes selected by machine learning, two targets, ESR2, and CDC42 were identified.\u003c/p\u003e\n \u003cp\u003eRNAseq data in the TPM format from the UCSC Xena database were processed using the Toil process. Data from normal tissues of COAD and GTEX from The TCGA were extracted, encompassing 308 normal and 290 tumor samples. RNAseq data were normalized to the log2 scale, facilitating expression comparisons (Fig. 12A). The Wilcoxon rank sum test indicated significantly reduced expression of CDC42 and ESR2 in tumor samples compared to normal samples. Pairing analysis between normal and tumor tissues from the same patient showed similar results (Figs. 12B-E). Survival analysis revealed that patients with lower quartile CDC42 expression in COAD exhibited lower survival probability than those with higher CDC42 expression, though no significant difference was observed for ESR2 (Figs. 12F, G). GSEA of CDC42identified multiple associated pathways, including allograft_rejection, epithelial-mesenchymal transition, inflammatory response, interferon_gamma_response, and myc_targets_v2 (Figs. 12H, I).\u003c/p\u003e\n \u003cp\u003eMolecular docking analysis was performed with CDC42 and 9 food contaminants. The lowest binding energies between these contaminants and the CDC42 protein were visualized. The contaminants interacted with key amino acid residues of CDC42, including THR-17, THR-58, ASP-57, ASP-11, ASP-176, LYS-16, ALA-13, ALA-176, ARG-186, GLY-15, GLU-178, and GLY-164. A binding energy of -7.47 kcal/mol indicated stable binding affinity between CDC42 and the contaminants (Fig. 13).\u003c/p\u003e\n\u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eThe pervasive presence of contaminants within the global food supply has sparked considerable apprehension regarding their detrimental effects on human health, particularly their propensity to trigger carcinogenesis\u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Although extensive research has been conducted on the toxicological profiles of food contaminants, such as Glyphosate, PFOS, Nitrosamines, and others\u003csup\u003e\u003cspan additionalcitationids=\"CR37\" citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e, the molecular pathways through which these agents promote the development of cancer remain inadequately elucidated. In this study, we harnessed a comprehensive array of advanced databases and bioinformatics network methodologies, including PubChem, ADEMTlab3.0, Protox3, GeneCards, OMIM, STRING, Machine Learning, and WGCNA, to identify and analyze relevant targets. Following this, we utilized Cytoscape software for an in-depth analysis of these targets. The DAVID database was then employed to perform GO and KEGG enrichment analyses on the core targets. Subsequently, we leveraged the PubChem database, in conjunction with AutoDock Vina, to conduct molecular docking studies,which facilitated an exploration of the intermolecular interactions between food contaminants and key target proteins. Collectively, our findings shed light on the potential molecular mechanisms by which food contaminants may contribute to the onset of carcinogenesis across various cancer types.\u003c/p\u003e \u003cp\u003eSince 1970, a wide range of environmental toxins, such as chemicals in tobacco smoke, air pollutants, and contaminants in food have been identified\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. These toxins are prevalent in daily life due to industrial growth and can impact the respiratory and cardiovascular systems and have been shown to be closely linked with the onset of various chronic diseases and cancers\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. Dietary exposure to these toxins poses significant health risks. For this study, we selected nine common dietary contaminants (Glyphosate, PFOS, Nitrosamines, PBDEs, Methylmercury, Dioxins, Acrylamide, Pyrrolizidine Alkaloids, and Aflatoxin) based on data from the National Institute of Environmental Health Sciences (NIEHS) and the Environmental Protection Agency (EPA). Our comprehensive analysis identified 69 targets intersecting with pan-cancer pathways, which were further refined through PPI analysis and GO/KEGG enrichment analyses. We focused specifically on BRCA, PRAD, and COAD cancers to elucidate potential mechanisms linking these contaminants to carcinogenesis.\u003c/p\u003e \u003cp\u003eIn the context of BRCA, our study identified 49 hub targets associated with contaminant-induced BRCA. The GO and KEGG enrichment analyses for these hub targets revealed a multitude of cancer-related functions and pathways, including the PI3K-Akt pathway, which is well-documented in the development and progression of breast cancer\u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. This finding is consistent with previous studies, such as the work by Kavarthapu R et al., which demonstrated that JUN, a key target identified in our study, acts as a critical regulator of the PI3K-Akt pathway and plays a significant role in breast cancer development\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. Similarly, EGFR, another key target identified in our study by machine learning algorithms and WGCNA, has been shown to drive tumor growth and metastasis in BRCA\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. Our RNA sequencing analysis, based on TCGA gene expression data, further supports these findings by revealing significantly lower expression of JUN and EGFR in tumor samples compared to normal tissues. Additionally, higher JUN expression was correlated with longer survival in BRCA patients, although no significant differences were observed for EGFR expression levels. Molecular docking analysis demonstrated stable binding affinities between the contaminants and JUN, suggesting direct interactions that may contribute to carcinogenesis. These results not only corroborate prior research but also provide novel insights into the molecular mechanisms underlying contaminant-induced BRCA, highlighting JUN as potential therapeutic targets.\u003c/p\u003e \u003cp\u003eConcerning PRAD, our research discovered 44 hub targets related to food contaminant-induced prostate cancer. The GO and KEGG enrichment analyses emphasized the considerable involvement of pathways, such as the PI3K-Akt pathway, which is widely documented in the progression of prostate cancer\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. This discovery is in line with recent studies that demonstrate the role of PI3K-Akt signaling in prostate cancer, particularly in regulating cell proliferation and apoptosis\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e. Specifically, our analysis identified MAPK14 and CDC42 as crucial targets, which have been associated with prostate cancer through their functions in inflammatory responses and DNA repair mechanisms. RNA sequencing analysis based on the TCGA data revealed a significantly decreased expression of MAPK14 in tumor samples compared to normal tissues, with lower MAPK14 expression being correlated with longer survival in PRAD patients. However, no significant disparities were observed in CDC42 expression levels between tumor and normal tissues from the same patient in the TCGA data, nor in the survival analysis. This led us to further validate MAPK14 as a potential therapeutic target. Molecular docking analysis demonstrated stable binding affinities between food contaminants and MAPK14, suggesting direct interactions that might contribute to carcinogenesis. This finding extends previous research by providing evidence of a molecular connection between environmental contaminants and key oncogenic pathways in prostate cancer.\u003c/p\u003e \u003cp\u003eFurthermore, 48 core targets were recognized to be associated with food contaminant-induced COAD, disclosing the significant involvement of cancer-related pathways such as transcriptional regulation and apoptosis. These discoveries are consistent with previous studies that have emphasized the crucial role of these pathways in the development of COAD\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e. Nevertheless, our study uniquely combines food contaminant targets with COAD-related genes, offering novel insights into potential environmental triggers. For example, our analysis identified CDC42 as a key target with significantly lower expression in tumor samples compared to normal tissues, and lower CDC42 expression was associated with worse survival outcomes. This is in line with recent studies indicating that the dysregulation of CDC42 is connected to aggressive cancer phenotypes \u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. In contrast, the expression of ESR2 did not demonstrate a significant correlation with survival, suggesting a more intricate role that requires further examination. Molecular docking analysis further disclosed stable binding affinities between food contaminants and CDC42, suggesting direct interactions that might contribute to carcinogenesis. These results not only support previous research but also introduce new potential therapeutic targets and biomarkers for COAD, emphasizing the significance of considering environmental factors in cancer development.\u003c/p\u003e \u003cp\u003eOur study innovatively integrates PPI analysis, machine learning algorithms, WGCNA, and molecular docking simulations to provide a comprehensive understanding of the molecular mechanisms underlying contaminant-induced carcinogenesis. This integrative approach not only identifies key targets and pathways significantly associated with food contaminants and cancer development but also reveals the intricate interactions between environmental factors and cellular processes. For instance, the identification of critical regulatory genes such as JUN and CDC42, which exhibit altered expression in tumor tissues and correlate with survival outcomes, offers novel insights for targeted therapy development. These findings could facilitate the design of small molecules or biologics that specifically inhibit the activity of these targets, thereby mitigating the carcinogenic effects of food contaminants.Moreover, our study highlights the potential for public health interventions. By pinpointing specific contaminants that interact with key cancer-related genes, our research provides a basis for developing policies aimed at reducing exposure to harmful substances in the food supply. This could significantly lower the incidence of contaminant-induced cancers and other related diseases, thereby enhancing overall public health. Additionally, the identified targets and pathways may serve as biomarkers for early detection and monitoring of cancer, further supporting personalized medicine approaches. Our study thus bridges the gap between environmental exposures and molecular mechanisms, offering both therapeutic and preventive strategies that could substantially impact cancer management and public health initiatives.\u003c/p\u003e \u003cp\u003eDespite the comprehensive nature of our study, several limitations should be acknowledged. First, our analysis relies heavily on in silico methods, and the identified targets and pathways need to be validated through experimental studies. Future work should focus on validating the molecular interactions identified through molecular docking simulations using in vitro and in vivo models. Additionally, the potential synergistic effects of multiple contaminants, which are often present in real-world scenarios, were not fully explored in this study. Future research should investigate the combined effects of contaminants on carcinogenesis, as this could provide more accurate insights into the risks posed by contaminated food.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eIn conclusion, our study provides novel insights into the molecular mechanisms linking food contaminants to carcinogenesis, identifying key targets and pathways that are significantly associated with contaminant-induced cancers. The findings highlight the potential of JUN, CDC42, and MAPK14 as therapeutic targets for mitigating the carcinogenic effects of food contaminants. While our study has several limitations, it lays the groundwork for future research aimed at validating these findings and exploring their potential applications in cancer prevention and treatment. Ultimately, a better understanding of the molecular underpinnings of contaminant-induced carcinogenesis could lead to more effective strategies for reducing the burden of cancer worldwide.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eCRediT authorship contribution statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBangsheng Chen:\u003c/strong\u003e Visualization, Conceptualization. \u003cstrong\u003eMaomao Li:\u003c/strong\u003eSoftware, Methodology. \u003cstrong\u003eWenzhu Lou:\u003c/strong\u003e Investigation.\u003cstrong\u003eZhiyan Wang:\u003c/strong\u003e Methodology, Investigation. \u003cstrong\u003eYi Gu:\u003c/strong\u003e Formal\u0026nbsp;analysis. \u003cstrong\u003eFeiyan Mao and Lian Tan:\u003c/strong\u003e Writing - review \u0026amp; editing, Data curation. \u003cstrong\u003eShuaishuai Huang:\u003c/strong\u003eWriting - original draft, Visualization, Methodology, Conceptualization.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u0026nbsp;\u003c/strong\u003eThis study was supported by the Natural Science Foundation of Ningbo, China (2022J039 to C.B.).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of Competing Interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eDeclaration of generative AI and AI-assisted technologies in the writing process.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDuring the preparation of this work, we used deepseek in order to improve the readability and language of the manuscript. After using this tool, we reviewed and edited the content as needed and take full responsibility for the content of the published article.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAll authors reviewed the manuscript\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eF. Bray, et al., Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries, CA Cancer J Clin 74(3) 229\u0026ndash;263 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK. Inamura, et al., Cancer as microenvironmental, systemic and environmental diseases: opportunity for transdisciplinary microbiomics science, Gut (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eV. Kayamba, P. Kelly, Environmental factors associated with gastric carcinogenesis, Curr Opin Gastroenterol 38(2) 156\u0026ndash;161 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Maher, A. Nowak, Chemical Contamination in Bread from Food Processing and Its Environmental Origin, Molecules 27(17) (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Sadighara, et al., Residues of carcinogenic pesticides in food: a systematic review, Rev Environ Health 39(4) 659\u0026ndash;666 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Masci, et al., Chromatographic Methods for the Determination of Glyphosate in Cereals Together with a Discussion of Its Occurrence, Accumulation, Fate, Degradation, and Regulatory Status, Methods Protoc 7(3) (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eT. Bonato, et al., Contamination of the terrestrial food chain by per- and polyfluoroalkyl substances (PFAS) and related human health risks: A systematic review, Sci Total Environ 961 178337 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eA. Vignesh, et al., Food contaminants: Impact of food processing, challenges and mitigation strategies for food security, Food Res Int 191 114739 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Sipos, et al., Physical and Chemical Methods for Reduction in Aflatoxin Content of Feed and Food, Toxins (Basel) 13(3) (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL. Zeng, et al., Mitochondrial Dysfunction-Molecular Mechanisms and Potential Treatment approaches of Hepatocellular Carcinoma, Mol Cell Biochem 480(4) 2131\u0026ndash;2142 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Sassano, et al., Exposure to per- and poly-fluoroalkyl substances and lung, head and neck, and thyroid cancer: A systematic review and meta-analysis, Environ Res 266 120606 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL. Bridgeman, et al., Effect of Acrylamide and Mycotoxins in SH-SY5Y Cells: A Review, Toxins (Basel) 16(2) (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Cheng, et al., Exploring the mechanism of PPCPs on human metabolic diseases based on network toxicology and molecular docking, Environ Int 196 109324 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eY. Li, et al., Air pollution and prostate cancer: Unraveling the connection through network toxicology and machine learning, Ecotoxicol Environ Saf 292 117966 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eN. He, et al., Elucidating the mechanism of plasticizers inducing breast cancer through network toxicology and molecular docking analysis, Ecotoxicol Environ Saf 284 116866 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZ. Xu, et al., Identification of molecular targets of Hypericum perforatum in blood for major depressive disorder: a machine-learning pharmacological study, Chin Med 19(1) 141 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP.C. Agu, et al., Molecular docking as a tool for the discovery of molecular targets of nutraceuticals in diseases management, Sci Rep 13(1) 13398 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM.A. Abbas, et al., Comprehensive understanding of the role of GPER in estrogen receptor-alpha negative breast cancer, J Steroid Biochem Mol Biol 241 106523 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB. Arjmand, et al., Molecular Docking as a Therapeutic Approach for Targeting Cancer Stem Cell Metabolic Processes, Front Pharmacol 13 768556 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Kim, et al., PubChem 2023 update, Nucleic Acids Research 51(D1) D1373-D1380%@ 0305\u0026ndash;1048 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL. Fu, et al., ADMETlab 3.0: an updated comprehensive online ADMET prediction platform enhanced with broader coverage, improved performance, API functionality and decision support, Nucleic Acids Res 52(W1) W422-W431 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Banerjee, et al., ProTox 3.0: a webserver for the prediction of toxicity of chemicals, Nucleic Acids Res 52(W1) W513-W520 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Kuhn, et al., STITCH: interaction networks of chemicals and proteins, Nucleic Acids Res 36(Database issue) D684-8 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD. Gfeller, et al., SwissTargetPrediction: a web server for target prediction of bioactive small molecules, Nucleic Acids Res 42(Web Server issue) W32-8 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eC. Chen, et al., SEA version 3.0: a comprehensive extension and update of the Super-Enhancer archive, Nucleic Acids Res 48(D1) D198-D203 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJ.S. Amberger, et al., OMIM.org: Online Mendelian Inheritance in Man (OMIM(R)), an online catalog of human genes and genetic disorders, Nucleic Acids Res 43(Database issue) D789-98 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eG. Stelzer, et al., The GeneCards Suite: From Gene Data Mining to Disease Genome Sequence Analyses, Curr Protoc Bioinformatics 54 1 30 1\u0026ndash;1 30 33 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD. Szklarczyk, et al., The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets, Nucleic Acids Res 49(D1) D605-D612 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Shannon, et al., Cytoscape: a software environment for integrated models of biomolecular interaction networks, Genome Res 13(11) 2498\u0026ndash;504 (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB.T. Sherman, et al., DAVID: a web server for functional enrichment analysis and functional annotation of gene lists (2021 update), Nucleic Acids Res 50(W1) W216-W221 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK. Yang, et al., Machine learning-optimized targeted detection of alternative splicing, Nucleic Acids Res 53(3) (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Langfelder, S. Horvath, WGCNA: an R package for weighted correlation network analysis, BMC Bioinformatics 9 559 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eC. UniProt, UniProt: the Universal Protein Knowledgebase in 2023, Nucleic Acids Res 51(D1) D523-D531 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Bugnon, et al., SwissDock 2024: major enhancements for small-molecule docking with Attracting Cavities and AutoDock Vina, Nucleic Acids Res 52(W1) W324-W332 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB.R. Rushing, M.I. Selim, Aflatoxin B1: A review on metabolism, toxicity, occurrence in food, occupational exposure, and detoxification methods, Food Chem Toxicol 124 81\u0026ndash;100 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD. Soares, et al., Glyphosate Use, Toxicity and Occurrence in Food, Foods 10(11) (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB. Zhang, et al., Impacts of PFOS, PFOA and their alternatives on the gut, intestinal barriers and gut-organ axis, Chemosphere 361 142461 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Alikord, et al., Food safety and quality assessment: comprehensive review and recent trends in the applications of ion mobility spectrometry (IMS), Crit Rev Food Sci Nutr 62(18) 4833\u0026ndash;4866 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL.B. Murrison, et al., Environmental exposures and mechanisms in allergy and asthma development, J Clin Invest 129(4) 1504\u0026ndash;1515 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eH. Orru, et al., The Interplay of Climate Change and Air Pollution on Health, Curr Environ Health Rep 4(4) 504\u0026ndash;513 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eI.M. Browne, et al., Optimal targeting of PI3K-AKT and mTOR in advanced oestrogen receptor-positive breast cancer, Lancet Oncol 25(4) e139-e151 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR. Kavarthapu, et al., Crosstalk between PRLR and EGFR/HER2 Signaling Pathways in Breast Cancer, Cancers (Basel) 13(18) (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eS. Sigismund, et al., Emerging functions of the EGFR in cancer, Mol Oncol 12(1) 3\u0026ndash;20 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eB.Y. Shorning, et al., The PI3K-AKT-mTOR Pathway and Prostate Cancer: At the Crossroads of AR, MAPK, and WNT Signaling, Int J Mol Sci 21(12) (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM. Hashemi, et al., Targeting PI3K/Akt signaling in prostate cancer therapy, J Cell Commun Signal 17(3) 423\u0026ndash;443 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJ. Bian, et al., Transcriptional Regulation of Wnt/beta-Catenin Pathway in Colorectal Cancer, Cells 9(9) (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eP. Zhou, et al., The Induction Mechanism of Ferroptosis, Necroptosis, and Pyroptosis in Inflammatory Bowel Disease, Colorectal Cancer, and Intestinal Injury, Biomolecules 13(5) (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eV. Justilien, et al., Oncogenic Ect2 signaling regulates rRNA synthesis in NSCLC, Small GTPases 10(5) 388\u0026ndash;394 (2019).\u003c/span\u003e\u003c/li\u003e \u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Food contamination, Cancer, Network Toxicology, Machine learning, Molecular docking","lastPublishedDoi":"10.21203/rs.3.rs-6625099/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6625099/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFood contamination is a significant global health threat, with carcinogenic potential, yet the molecular pathways linking contaminants to cancer remain poorly understood. This study aimed to identify key molecular targets mediating the carcinogenic effects of food contaminants. We utilized multiple online databases to identify target genes associated with nine prevalent dietary contaminants (Glyphosate, Perfluorooctane sulfonate, Nitrosamines, Pentabromodiphenyl ethers, Methylmercury, Dioxins, Acrylamide, Pyrrolizidine Alkaloids, and Aflatoxin) and pan-cancer. Protein-protein interaction (PPI) analysis and visualization were conducted on intersecting genes, followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) functional enrichment analyses to uncover potential mechanisms. We focused on breast (BRCA), prostate (PRAD), and colon (COAD) carcinomas due to their significant pathway associations. Hub genes were prioritized using an integrative strategy combining topological algorithms in Cytoscape (Centiscape, MCODE, and cytohubba's MCC), machine learning validation, and Weighted Gene Co-expression Network Analysis (WGCNA). Molecular docking simulations were performed to examine interactions between contaminants and hub genes. We identified 69 pan-cancer-intersected targets. Comprehensive enrichment analyses revealed significant cancer-associated pathways. Hub gene prioritization identified JUN in BRCA, CDC42 in COAD, and MAPK14 in PRAD as critical regulatory targets. Validation using The Cancer Genome Atlas (TCGA) data confirmed statistically significant differential expression patterns (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) for these targets across respective malignancies. Gene Set Enrichment Analysis (GSEA) delineated pathway activation profiles consistent with tumor progression mechanisms. Molecular docking simulations demonstrated robust binding affinities (binding energy \u0026le;-5.0 kcal/mol) between contaminants and structural domains of identified hub targets, suggesting direct mechanistic interactions. Our study elucidates the molecular mechanisms underlying dietary carcinogens, identifies potential therapeutic targets, and highlights the need for enhanced food safety policies. This integrative approach combining molecular and clinical insights may inform precision public health interventions.\u003c/p\u003e","manuscriptTitle":"Unraveling the Carcinogenic Mechanisms of Food contaminants through Network Toxicology, Machine Learning, and Molecular Docking","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-05-28 11:07:45","doi":"10.21203/rs.3.rs-6625099/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"10110936-9b8f-4aa5-b664-344e6193dc4b","owner":[],"postedDate":"May 28th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":48981790,"name":"Biological sciences/Cancer"},{"id":48981791,"name":"Earth and environmental sciences/Environmental social sciences"},{"id":48981792,"name":"Health sciences/Health care"}],"tags":[],"updatedAt":"2025-07-21T17:23:33+00:00","versionOfRecord":[],"versionCreatedAt":"2025-05-28 11:07:45","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6625099","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6625099","identity":"rs-6625099","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}