LipiDecipher: A Structure-Oriented Framework for Mechanistic Interpretation in Clinical Lipidomics

LipiDecipher: A Structure-Oriented Framework for Mechanistic Interpretation in Clinical Lipidomics | 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 Method Article LipiDecipher: A Structure-Oriented Framework for Mechanistic Interpretation in Clinical Lipidomics Anliang Huang, Yunshu Zhang, Tingting Bai, Xiaoyang Yuan, Dong Shang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7238340/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 LipiDecipher is a structure-oriented framework designed to overcome the functional interpretation bottleneck in lipidomics. By integrating lipid-protein-pathway mapping with a unique lipid deconstruction algorithm, it connects molecular structure to biological function. Analyzing a myocardial infarction cohort, LipiDecipher revealed novel mechanisms, including a progressive depletion of phosphatidylinositol-associated fatty acids in recurrent patients, which points to deteriorating membrane repair capacity. This powerful paradigm enables the discovery of novel biological insights and robust biomarkers, significantly advancing clinical lipidomics research. Lipidomics Myocardial Infarction Computational Biology Bioinformatics Structure-Function Relationship Lipid Deconstruction Biomarker Discovery Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Background Lipids are fundamental biological molecules essential for cellular structure, energy homeostasis, and signal transduction. While conventional clinical assays measuring total triglycerides and cholesterol serve as primary predictors for cardiovascular events, their diagnostic specificity and early-prediction sensitivity are limited[ 1 , 2 ]. The advent of high-throughput mass spectrometry has propelled lipidomics to the forefront of clinical research, offering a comprehensive and specific landscape of circulating lipids that can significantly enhance prognostic accuracy for complex diseases[ 3 ]. In the post-genomic era, clinical lipidomics, driven by specific clinical questions and the complexity of patient samples, promises to revolutionize early diagnosis, precision medicine, and therapeutic strategies[ 4 – 7 ]. Despite our ability to measure thousands of lipid species, the sheer diversity and structural complexity of the lipidome present formidable analytical and interpretive challenges. Current research is often confined to analyzing individual lipid classes, failing to capture the global, systems-level picture. In other omics disciplines, functional interpretation is greatly facilitated by well-established frameworks like Gene Ontology[ 8 ] (GO) and pathway enrichment analysis (e.g., Kyoto Encyclopedia of Genes and Genomes[ 9 – 11 ] (KEGG), which effectively translate vast datasets into biological insights. However, the lipidomics field currently lacks a comparably mature analytical paradigm, creating a significant bottleneck. This gap stems from a fundamental mismatch between generic analytical logic and the unique nature of lipids. The biological function of a lipid is intrinsically tied to its detailed chemical structure, yet most analytical workflows, particularly those adapted from general metabolomics, fail to account for this. They often overlook the critical fact that subtle structural variations—such as differences in fatty acyl chain length or the number of double bonds—can lead to vastly different, or even opposing, biological activities within the same lipid class. This limitation often restricts functional analysis to a superficial level, masking the precise structure-driven mechanisms at play. To unlock the full clinical potential of lipidomics, a new analytical strategy is urgently needed—one that can systematically link lipid structural features to biological function and enable a more profound, mechanistically-informed interpretation. To address this challenge, we developed LipiDecipher, a comprehensive and modular analytical framework designed to navigate the complexities of lipidomics data. The entire workflow, from data acquisition to biological interpretation, is schematically illustrated in Fig. 1 . LipiDecipher integrates multi-scale statistical analyses with a structure-centric approach, organized into a dual-pronged analytical strategy to provide deeper biological insights. The first arm of this strategy, Workflow 1: Intact Lipid-Level Analysis, focuses on the intact lipid molecule level. It builds upon traditional statistical analyses by applying a "lipid-protein-pathway" mapping to features of interest for functional enrichment. This method links identified lipid structures to their known protein targets and subsequently to established biological pathways, aiming to provide a more precise and mechanistically grounded interpretation than conventional annotation-based methods. The second arm, Workflow 2: Structurally-Resolved Analysis, delves into the sub-molecular level. Its core feature is a lipid deconstruction module that algorithmically infers deep structural properties—such as fatty acyl composition—from standard LC-MS/MS data. This capability enables the calculation of enzyme activity proxies and facilitates the direct correlation of these granular features with clinical phenotypes. Through this dual-workflow structure, LipiDecipher is presented as a reproducible and extensible analytical paradigm. To promote its adoption and impact, we provide a detailed case study and open-source R code. By doing so, we aim to empower the research community to discover novel biological mechanisms and robust biomarkers, thereby accelerating the translation of lipidomics findings into clinical practice. Results Global Landscape of the Lipidome during Myocardial Infarction Progression To demonstrate the analytical power of the LipiDecipher framework in a real-world clinical scenario, we performed untargeted lipidomics analysis on samples from three clinical groups: Healthy Controls (HC, n = 50), Acute Myocardial Infarction (MI, AMI, n = 50), and Post-PCI Recurrent Myocardial Infarction (PRMI, n = 35). Before delving into biological differences, we first assessed the stability of our analytical workflow. As shown in the alluvial plot (Fig. 2 A), all Quality Control (QC) samples exhibited a highly consistent lipid composition pattern, with remarkably similar widths and color proportions of their internal flux bands. This provides strong evidence for the stability and reproducibility of the analytical process, laying a solid foundation for subsequent biological insights. First, we utilized a global heatmap (Fig. 2 B) to reveal overall patterns of change. This plot displays the expression profiles of all lipid molecules, clearly revealing systematic lipid remodeling during disease progression. Compared to the HC group, the AMI and PRMI groups presented multiple clusters of significantly upregulated (brown modules) or downregulated (blue modules) lipids, indicating a global alteration in lipid metabolism. Next, to quantify this change at the compositional level, we employed an alluvial plot (Fig. 2 B) to analyze the compositional shift of major lipid categories. Compared to the HC group (where glycerolipids (GL) constituted ~ 38.9% and glycerophospholipids (GP) ~ 54.1%), the relative abundance of GL increased in the AMI and PRMI groups (to ~ 40.9% and ~ 45.5%, respectively), while the relative abundance of GP correspondingly decreased (to 51.6% and 47.2%). This result suggests that the onset and recurrence of MI may be associated with a disrupted balance between lipid energy storage and cell membrane structure. To precisely identify the key lipid main classes driving this remodeling and to elucidate their hierarchical relationships, we constructed a composite plot combining a Sankey diagram with lollipop charts (Fig. 2 C). The Sankey diagram clearly illustrates the hierarchical composition of the untargeted lipidomics data, showing the distribution from major categories to main classes. The width of the flows is proportional to the number of unique lipid molecules within each branch, clearly revealing the primary constituent classes within each category. For instance, the GL category is predominantly composed of triglycerides (TG), while the GP category mainly comprises phosphatidylcholines (PC) and phosphatidylethanolamines (PE). Corresponding to this structure, the flanking lollipop charts quantify the abundance changes at both the category and main class levels. At the category level, GL and sterol lipids (ST) were upregulated in the AMI and PRMI groups compared to HCs, whereas GP and sphingolipids (SP) showed a downward trend. At the more granular main class level, we observed that the trends of change in the AMI (triangles) and PRMI (circles) groups were highly consistent across most main classes. TGs, cholesterol esters (ChE) as major forms of energy storage, and the signaling molecule diacylglycerol (DG) all showed significant and consistent upregulation (log₂FC range approx. [0.4 to 1]). Conversely, several key phospholipids constituting cell membranes, including PC, PE, and phosphatidylinositols (PI), exhibited significant downregulation (log₂FC range approx. [-0.2 to -0.5]). This strongly implies that the lipid dysregulation pattern established after the acute event persists in recurrent patients. However, subtle differences revealed deeper information: compared to the AMI group, TG, DG, and PI showed more substantial changes in the PRMI group, suggesting that the persistent, high-level dysregulation of these specific lipids may be linked to the mechanisms of MI recurrence. Category-Specific LDA for Identifying Efficient Discriminatory Features Following the global analysis that revealed overall trends, the first major branch of our analytical paradigm—Intact Lipid-Level Analysis—aims to deeply investigate the biological significance of intact lipid molecules from three complementary perspectives. First, from a "static-discriminatory" viewpoint, we sought to precisely identify the key features that contribute most to distinguishing between the different clinical groups from the complex pool of lipid molecules. To this end, we employed a category-specific Linear Discriminant Analysis (LDA) method (Figure S1 ). This approach does not perform a single global analysis on all lipids but instead constructs an independent classification model for each major lipid category, thereby enabling a more refined elucidation of the distinct roles of different lipid classes in the pathophysiology of the disease. Figure 3 centralizes the LDA results for four major lipid categories: ST, SP, GP, and GL. The central scatter plot in each sub-figure shows the projection of samples onto the two main discriminant axes (LD1 and LD2). Through this module, we not only achieved effective group separation but also assigned clear biological meaning to these mathematically constructed discriminant axes. A consistent pattern across all models was that the LD1 axis primarily captured the differences between HC and the disease states (AMI/PRMI), while the LD2 axis mainly distinguished between the two distinct disease stages, acute MI (AMI) and PRMI. The peripheral bar plots quantify and display the lipid molecules that contribute most to the classification (i.e., those with the highest loadings). This feature is central to LipiDecipher's "difference attribution" capability. Through this module, we can clearly identify the key molecules driving different biological comparisons. In the ST category (Fig. 3 A), although there was slightly more overlap between groups, the model (LD1 contribution 95.74%) could still effectively distinguish the HC and PRMI groups, driven primarily by the downregulation of various cholesterol esters (ChE), such as ChE (20:5) and ChE (22:6). In the SP category (Fig. 3 B), the distinction between "health vs. disease" (LD1 contribution 88.43%) relied mainly on various long-chain SM and ceramides (Cer). Specifically, the upregulation of molecules like SM (d42:1) and Cer (d41:1) was a significant feature for distinguishing the AMI/PRMI groups. In the GP category (Fig. 3 C), the key drivers for distinguishing "health vs. disease" (primarily reflected by the LD1 axis, 85.64% contribution) were mainly the downregulation of various polyunsaturated fatty acid (PUFA)-containing phospholipids, such as PC (38:6), PC (38:4), and PI (34:1). In the GL category (Fig. 3 D), the discriminatory power was almost entirely dominated by various TGand diacylglycerols (DG). For instance, TG (54:3) and DG (36:3) were significantly elevated in the disease groups, becoming core markers for distinguishing them from the HC group. The category-specific LDA analysis not only constructed high-precision classifiers for each lipid category (cross-validation showed an average accuracy > 95%, Figure S1 ) but, more importantly, successfully identified key lipid molecules with high discriminatory power from a classification perspective, thus laying a foundation for more rigorous differential quantification and functional exploration. Differential Comparison and Functional Mapping: Quantifying Changes at Critical Stages After LDA provided an overview of key discriminatory features, we next proceeded from a "pairwise-comparison" perspective. We employed rigorous statistical tests to precisely quantify the lipids that changed significantly during the critical stages of disease "onset" (AMI vs. HC) and "progression" (PRMI vs. AMI) (Fig. 4 A). However, accurately identifying differential lipids is only the first step. We immediately faced a core challenge in lipidomics analysis: how to link these specific lipid molecules to concrete biological pathways. Traditional enrichment analysis tools are not directly applicable to lipids, which greatly limits our understanding of the mechanisms behind lipid remodeling. To address this challenge, our LipiDecipher analytical paradigm applies an innovative "lipid-protein-pathway" mapping strategy. This strategy serves as a critical bridge connecting lipid structure to biological function. Using the differential lipids most closely associated with MI recurrence identified in the "PRMI vs. AMI" comparison as input, we first mapped them to known protein targets (Uniprot ID) via their precise chemical structures (e.g., SMILES format) using the SwissLipids and LipidMaps databases. This step successfully translated changes at the lipid molecule level into information at the protein level. Based on the resulting set of protein targets, we could then perform standard pathway enrichment analysis using the well-established clusterProfiler tool. The results clearly revealed the biological processes in which these recurrence-associated differential lipids are involved:(1) KEGG pathway enrichment analysis (Fig. 4 B) showed that the protein targets of these lipids were significantly enriched in core lipid pathways such as Glycerophospholipid metabolism, Sphingolipid metabolism, and Arachidonic acid metabolism. (2) GO enrichment analysis (Fig. 4 C) provided more detailed functional annotations. At the Biological Process level, the protein targets were highly associated with phospholipid metabolism and fatty acid metabolism. At the Molecular Function level, they pointed mainly to acyltransferase activity. At the Cellular Component level, they were closely related to the peroxisome and lipid droplet. In summary, our proposed analytical workflow not only precisely identified lipid molecules associated with disease recurrence but, more importantly, successfully linked these molecules to specific biological pathways through a structure-driven "lipid-protein-pathway" mapping strategy. This revealed that MI recurrence may be associated with inflammation and energy metabolism disorders, fully demonstrating the power of our analytical paradigm in bridging the critical gap from lipid structure to functional insight. Dynamic Trend Clustering: Delineating Co-regulated Functional Modules Although pairwise differential analysis can pinpoint changes at specific stages, it cannot fully depict the continuous dynamic trajectories of lipid molecules across the entire disease spectrum. Therefore, as the final step from a "dynamic-panoramic" perspective, we employed Fuzzy C-Means (FCM) clustering analysis. This was aimed at identifying co-regulated functional modules of lipids from the panoramic view of "health-acute-recurrence". This analysis partitioned all lipid molecules into five functional modules with distinct dynamic trends (Fig. 5 A, left panels). Each module represents a specific biological response pattern: Cluster 3 (Sustained Upregulation): Lipids were significantly elevated during the AMI stage and remained high or increased further during the PRMI stage. Cluster 5 (Sustained Downregulation): Lipids continuously decreased after the onset of the disease.Cluster 1 (V-shaped - Down then Up): Lipids acutely decreased during the AMI stage but showed a recovery trend in the PRMI stage.Cluster 2 (Acute Upregulation): Lipids sharply increased during the AMI stage but tended to fall back in the PRMI stage.Cluster 4 (Moderate Downregulation): Exhibited a milder downward trend. The true value of these dynamic modules lies in our ability to assign clear biological functions to each dynamic pattern through our core "lipid-protein-pathway" mapping strategy. By performing independent pathway enrichment analysis on each cluster module (Fig. 5 A, right panel), we found that lipids with different dynamic patterns participated in distinctly different biological processes: For example, the protein targets of lipids in the sustainedly upregulated Cluster 3 were significantly enriched in the Phosphatidylinositol signaling system and glycerolipid metabolic process, pointing to signaling and metabolic remodeling that are continuously activated during disease progression. In stark contrast, lipids in the sustainedly downregulated Cluster 5 were associated with lipase activity and the phosphatidylinositol 3-kinase complex, suggesting the inhibition of certain key signaling pathways. To further reveal the molecular composition of these functional modules, we constructed a bipartite network graph connecting lipids to their clusters (Fig. 5 B). This network clearly displays the lipid main class composition within each dynamic module. For instance, the network visually demonstrates that the sustainedly upregulated Cluster 3 is primarily composed of TG and PC, while the sustainedly downregulated Cluster 5 is rich in Cer and TG. This analysis not only identified co-regulated groups of lipids but also tightly linked specific dynamic patterns, functional pathways, and the chemical classes of lipids. Structurally-Resolved Analysis: Deconstructing Lipids to Reveal Systematic Metabolic Remodeling Having completed the multi-dimensional analysis of intact lipids, we initiated the second major branch of our analytical paradigm—Structurally-Resolved Analysis. This was designed to transcend the traditional approach of treating lipids as independent entities and to delve into the sub-molecular structural level. The core of this strategy is Lipid Deconstruction, an algorithm that proportionally attributes the abundance of each parent lipid to its constituent fatty acyl chains. Notably, these fatty acyl chains originate from relatively stable complex lipid pools (e.g., TG, PC, SM), and their abundance changes are more indicative of long-term metabolic adaptation and remodeling rather than the transient fluctuations of free fatty acids in the blood. After applying the deconstruction algorithm, we first constructed a global heatmap (Fig. 6 ) to visually present the overall abundance patterns of fatty acyl chains across all samples. In this plot, each row represents a unique fatty acyl chain, and each column represents a biological sample, with color intensity reflecting its standardized abundance. Through the side row annotations, readers can quickly identify the structural features (e.g., saturation, chain length) of each fatty acyl chain and its primary lipid class of origin. This map provides a macroscopic, structured data landscape for subsequent quantitative analysis and offers initial hints of systematic differences between the clinical groups. To examine these differences in greater detail, we further calculated the fold change of each fatty acyl chain relative to the HC group within each major lipid class (Fig. 7 A). This plot revealed complex, class-specific patterns, highlighting the necessity of analyzing fatty acyl chains on different lipid backbones independently. (1) Heterogeneous changes in the GP family: We observed significant heterogeneity within the GP family. In contrast to the downward or stable trend of most fatty acyl chains in PC and PE, those in lysophosphatidylcholines (LysoPC) showed a moderate and general upregulation. Particularly noteworthy were the fatty acids associated with PI. Although their changes were relatively modest, they were consistently downregulated in the disease groups, with a significantly greater decrease in the recurrent group (PRMI, circles) than in the acute group (AMI, triangles), presenting a "dose-effect" pattern possibly related to disease severity or progression. (2) Unique patterns within TG: Although TGs were generally upregulated, this plot also revealed some unique molecular changes. For example, the saturated fatty acid TG (10:0) showed a sharp, nearly 4-fold increase in the AMI group but returned to near-control levels in the PRMI group. This suggests that TG (10:0) might be a transiently changing molecule closely associated with the acute-phase stress response, rather than a persistent marker. (3) Trends in other classes: Meanwhile, fatty acyl chains in Cer, ChE, and DG also exhibited a general upward trend similar to TGs, especially in the polyunsaturated fraction. The changes in SM were more complex and showed some saturation specificity. Collectively, this fine-grained, class-specific analysis not only confirms that fatty acid metabolism on different lipid backbones is differentially regulated but also successfully identifies key fatty acyl chains with unique dynamic behaviors (such as the progressive decrease of PI and the acute increase of TG (10:0), providing more precise targets for subsequent mechanistic investigation and biomarker discovery. To systematically test the overall trends of fatty acid elongation and desaturation on a global scale, we performed a structure-abundance correlation analysis. This analysis quantifies the linear relationship between the structural parameters of fatty acyl chains (carbon number and double bond count) and their abundance changes (log₂FC). The results showed that in the "AMI vs. HC" and "PRMI vs. HC" comparisons, the carbon chain length of fatty acids was positively correlated with their log₂FC values (Fig. 7 B). This finding is consistent with the activation of the overall fatty acid elongation process. The number of double bonds, however, did not show a clear correlation, suggesting that the desaturation process may be less strongly associated with the disease state. However, correlation analysis can only reflect overall trends and cannot directly point to the activity changes of specific enzymes. Therefore, we finally calculated specific product-to-substrate ratios as proxies for enzyme activity to directly infer changes in key metabolic enzymes. The results (Fig. 7 C, D) showed that compared to the HC group, the activity proxy for stearoyl-CoA desaturase (SCD-18, 18:1/18:0) was significantly elevated in both the AMI and PRMI groups (p < 0.001). Although the activity proxies for elongase (ELOVL) and other desaturases (SCD-16, FADS1, FADS2) did not show statistically significant differences between groups, the clear change in SCD-18 activity confirms that even within the stable complex lipid pool, the activation of the desaturation pathway is a detectable key event in the pathophysiology of MI. In summary, by integrating lipid deconstruction, class-specific quantification, structure-abundance correlation, and enzyme activity proxy analysis, our analytical paradigm successfully revealed the systematic remodeling of fatty acid metabolism in myocardial infarction from a structural level. This provides deeper, structure-driven insights for understanding the mechanisms of the disease. With this, we have completed a comprehensive exploration from the global landscape to intact molecules and down to sub-molecular structures, laying a solid foundation for the subsequent discussion. Discussion In this study, we utilized the LipiDecipher analytical framework to conduct a systematic, multi-dimensional analysis of lipidomics data related to myocardial infarction. This application demonstrated the framework's robust capabilities in processing complex lipidomics data, identifying potential biomarkers, and elucidating disease-associated biological mechanisms. More importantly, it offers a practical solution for the paradigm shift in lipidomics research, moving from molecular identification to functional interpretation. LipiDecipher: A Synergistic Strategy Integrating Multi-Source Information and Deconstructing Lipid Structure Accurate lipid identification and annotation serve as the critical link between raw data and biological meaning in lipidomics research. However, the inherent limitations of single databases and the structural complexity of lipids often lead to incomplete or inaccurate annotations[ 12 , 13 ]. LipiDecipher addresses these challenges through two core designs. Firstly, by integrating multiple authoritative databases[ 14 – 16 ] (e.g., Lipid Maps, SwissLipids), LipiDecipher builds a more comprehensive and robust annotation system. This integration is not merely a data aggregation but a strategy of cross-validation and information complementarity. It significantly enhances annotation accuracy and coverage, effectively mitigating the "information silo" problem caused by database discrepancies and providing a high-quality, structured data foundation for subsequent lipid-protein-pathway mapping. Secondly, a hallmark feature of LipiDecipher is its lipid structure deconstruction capability. Whereas traditional analysis often stops at identifying intact molecules like TG (54:3), our framework delves deeper by deconstructing them into their constituent fatty acyl chains. As demonstrated in our analysis of Fig. 7 , this ability to move from macroscopic molecules to microscopic structures allows for a systematic evaluation of how fatty acids with different chain lengths and saturation levels change during disease progression within a class-specific context. This unique analytical dimension enables LipiDecipher to uncover the deeper, structure-driven mechanisms hidden beneath the changes in intact lipid molecules. Systematic Lipid Remodeling: From Global Landscape to Dynamic Modules Our investigation began at a global level, revealing a profound and systematic lipid remodeling during MI and its recurrence. A salient feature was the fundamental shift in lipid composition from healthy controls to AMI and PRMI patients: a progressive increase in the relative abundance of glycerolipids (GL) for energy storage, with a concomitant decrease in glycerophospholipids (GP), the structural backbone of cell membranes (Fig. 2 B). This finding established a key theme for our study—that the pathophysiology of MI involves a dysregulation of the balance between energy metabolism and cell membrane homeostasis. To comprehensively capture the key molecules driving this shift, we employed a multi-dimensional screening strategy: Static Discrimination (LDA, Fig. 3 ), from a classification perspective, efficiently identified the most discriminating lipids. For instance, the downregulation of various PUFA-containing phospholipids (e.g., PC (38:6), PI (34:1)) and the upregulation of long-chain sphingolipids (e.g., SM(d42:1), Cer(d41:1)) were core drivers distinguishing the 'healthy vs. disease' states. Dynamic Clustering (FCM, Fig. 5 ), from a panoramic disease perspective, revealed co-regulated lipid modules. For example, the 'Sustained Upregulation' Cluster 3, primarily composed of triglycerides (TG) and phosphatidylcholines (PC), was functionally enriched in the "Phosphatidylinositol signaling system," suggesting its continuous activation. Conversely, the 'Sustained Downregulation' Cluster 5, rich in ceramides (Cer), was associated with the inhibition of "lipase activity." Through this progressive analysis, from global composition to static discrimination and dynamic patterns, we cross-validated the central role of key lipid classes such as TGs, PCs, and Cers in the pathology of MI, laying a solid foundation for subsequent mechanistic exploration. "Lipid-Protein-Pathway" Mapping: Overcoming the Hurdle of Functional Enrichment in Lipidomics One of the major bottlenecks in current lipidomics research is linking identified differential lipids to specific biological pathways[ 17 , 18 ]. To address this challenge, we proposed an innovative "lipid-protein-pathway" mapping strategy. By predicting the protein targets of differential lipids using databases like SwissLipids[ 16 ], we successfully shifted the analytical dimension from lipids to proteins, thereby enabling the use of well-established pathway enrichment tools. In our study (Fig. 4 ), we applied this strategy to focus on the differential lipids most closely associated with MI recurrence (PRMI vs. AMI). Their protein targets were successfully mapped to core pathways, including Glycerophospholipid metabolism, Sphingolipid metabolism, and Arachidonic acid metabolism. These findings are highly consistent with existing knowledge: dysregulation of glycerophospholipid metabolism is directly linked to cardiomyocyte injury and dysfunction[ 19 ], the role of sphingolipids (especially ceramides) in apoptosis and inflammation makes them a focal point in MI research[ 20 , 21 ], and arachidonic acid metabolites, as key inflammatory mediators, play an undisputed role in MI pathophysiology[ 22 , 23 ]. Furthermore, GO enrichment analysis pointed to cellular components such as lipid droplets and peroxisomes, forming a logical loop with our observations of altered energy metabolism (TG accumulation) and potential oxidative stress at the global level[ 24 – 26 ]. These results not only validate the effectiveness of our mapping strategy but also provide a new, systems-biology perspective for understanding the mechanisms of MI recurrence. Structural Deconstruction: Revealing Deeper Metabolic Mechanisms from Fatty Acyl Chain Remodeling Through LipiDecipher's unique deconstruction analysis, we were able to transcend the analysis of intact lipids and directly investigate the association between fatty acyl chain structure and disease state. Importantly, our analysis is based on stable, complex lipid pools, making our findings more reflective of long-term metabolic remodeling Our analysis revealed complex and biologically meaningful patterns. First, the structure-abundance correlation analysis in Fig. 7 B showed that the carbon chain length of fatty acids was positively correlated with their abundance changes, suggesting that the fatty acid elongation process is systematically activated during the disease. However, the number of double bonds showed no clear correlation, indicating that the desaturation process might be more complex. To investigate this complexity, we calculated proxies for enzyme activity (Fig. 7 C). The results showed that the activity of stearoyl-CoA desaturase (SCD-18) was significantly elevated, which aligns with existing literature on the role of SCD1 in cardiovascular protection and anti-inflammation[ 27 , 28 ]. In contrast, other desaturases (FADS1/2) showed no significant changes. This suggests that, against a background of unclear overall desaturation trends, the SCD-18 pathway is specifically activated and may be a key metabolic node driving disease progression. Notably, the class-specific deconstruction analysis (Fig. 7 A) revealed fine-grained regulatory mechanisms unattainable by traditional methods: Progressive Deterioration Pattern: Our deconstruction analysis found that fatty acyl chains associated with PI exhibited a "dose-effect" pattern, with a significantly greater decrease in the PRMI group than in the AMI group. This pattern strongly suggests a continuous deterioration of cell membrane signaling and repair capacity as the disease progresses. The biological basis for this inference lies in the central role of PI and its phosphorylated derivatives as key nodes in cellular signaling networks. For instance, the critical signaling molecule PtdIns(3,4,5) P₃ regulates cell survival, proliferation, and response to mechanical stress by activating downstream proteins—processes crucial for cardiomyocyte repair after ischemic injury[ 29 , 30 ]. Thus, the unique insight from LipiDecipher is that it not only confirms the central role of the PI signaling pathway in MI pathology but also reveals for the first time that this dysfunction may be linked to the progressive depletion of the PI molecular pool itself (specifically its fatty acyl chains), offering a deeper, structure-driven perspective on the mechanisms of disease exacerbation. Acute Stress Pattern: The short-chain saturated fatty acid TG (10:0) showed a sharp, nearly 4-fold increase exclusively in the AMI group, returning to baseline in the PRMI group. This strongly suggests that it is a transient marker closely associated with the acute-phase stress response, rather than a persistent marker of recurrence. These findings fully demonstrate the unique value of deconstruction analysis in revealing fine-grained, structure-driven regulatory mechanisms. Significance and Limitations The core contribution of this study is that we not only revealed a series of profound lipid metabolic disturbances during MI and its recurrence but, more importantly, we proposed and validated a systematic, structure-oriented lipidomics analysis paradigm named LipiDecipher. Conceptually, it advances lipidomics research from "discovering differential molecules" to "understanding structure-function relationships." Technically, through its "multi-dimensional feature screening" and "lipid-protein-pathway" mapping, it provides a viable approach to address the long-standing challenges of biomarker reliability and functional annotation in the field. However, it is important to acknowledge several limitations. First, as previously mentioned, all protein targets and pathways predicted bioinformatically require subsequent experimental validation. Second, this study's cross-sectional design reveals association, not causation; future longitudinal cohort studies are warranted to track the dynamic changes of lipids and validate their prognostic value. Finally, our findings are based on a specific clinical cohort, and their generalizability needs to be tested in larger, more diverse populations. Conclusion By applying the LipiDecipher analytical framework, this study systematically delineated the multi-level lipid remodeling landscape during the onset and recurrence of myocardial infarction and successfully linked specific lipid structural changes to key metabolic pathways. Our proposed framework demonstrates its potential in handling complex lipidomics data, identifying reliable biomarkers, and revealing disease-related biological mechanisms. We believe that LipiDecipher, as an open-source and extensible tool, will help advance clinical lipidomics research and accelerate the translation of lipidomics findings into novel strategies for improving cardiovascular disease diagnosis and treatment. The R code for the LipiDecipher framework is publicly available on GitHub at: https://github.com/AaronHwang8720/LipiDecipher . Methods Clinical Cohort and Sample Information This study enrolled a clinical cohort comprising three groups of subjects: a healthy control group (HC, n = 50), an acute myocardial infarction group (AMI, n = 50), and a group of patients with post-percutaneous coronary intervention (PCI) recurrent myocardial infarction (PRMI, n = 35). Detailed clinical baseline characteristics of the cohort are provided in Supplementary Table S1 . UHPLC-MS/MS Lipidomics Analysis Sample Preparation and Lipid Extraction Lipid extraction was performed using a methyl tert-butyl ether (MTBE) liquid-liquid extraction method. The core extraction procedure was as follows: 120 µL of methanol containing internal standards was added to 20 µL of serum, and the mixture was vortexed (1000 rpm, 5 min). Subsequently, 360 µL of MTBE and 100 µL of ultrapure water were added to induce phase separation, followed by another vortexing step (1000 rpm, 10 min) and incubation at room temperature for 10 minutes. After centrifugation at 13,000 g for 15 min at 4°C, the upper organic phase was collected and lyophilized. Prior to analysis, the lipid residue was reconstituted in 80 µL of acetonitrile-isopropanol (1:1, v/v). Quality control (QC) samples were prepared by pooling equal aliquots from all biological samples and were processed alongside them. Untargeted Lipidomics Analysis Untargeted analysis was conducted on a UPLC-HRMS system, consisting of a Thermo Scientific Ultimate™ 3000 UPLC system coupled to a Thermo Scientific Q Exactive™ Quadrupole-Orbitrap high-resolution mass spectrometer (Thermo Scientific, USA). Chromatographic separation was achieved using an Accucore C30 core-shell column. The mobile phase consisted of (A) 60% acetonitrile in water and (B) 90% isopropanol/10% acetonitrile, both containing ammonium formate and 0.1% formic acid. The gradient elution program was as follows: initial 10% B, ramped linearly to 50% B in 5 min, then further ramped to 100% B over 23 min. The mass spectrometer was equipped with a heated electrospray ionization (H-ESI) source, and data were acquired in both positive and negative ionization modes. Ionization parameters were set as follows: sheath gas flow rate, 45 arb; auxiliary gas flow rate, 10 arb; capillary temperature, 320°C; heater temperature, 355°C; S-Lens RF level, 55. Data were collected in full scan mode. Lipidomics Data Preprocessing Raw mass spectrometry data underwent a series of standardized preprocessing steps. In brief, the data were first subjected to quality control (QC), where lipid features were filtered out if they had a missing value rate > 50% in QC samples or > 20% in biological samples. Unstable features with a coefficient of variation (CV) > 30% in QC samples were also removed. Subsequently, remaining missing values were imputed using the k-Nearest Neighbors (kNN) algorithm. Finally, all lipids were annotated, assigned unique lipid IDs, and classified into different Categories and Main Classes based on their chemical structures. LipiDecipher: A Structure-Oriented Lipidomics Analysis Framework To extract structure-function information from complex lipidomics data, we developed and applied the LipiDecipher analysis framework. This framework includes the following core modules: Lipid Deconstruction and Abundance Attribution Traditional lipid analysis treats each lipid molecule as an independent entity. To overcome this limitation, we employed a "deconstruction" strategy. This algorithmic approach proportionally attributes the instrumental response (abundance) of each parent lipid to its constituent fatty acyl chains based on their molecular weight relative to the total molecular weight of the parent lipid. First, for a parent lipid ( L ), the attributed abundance of a single fatty acyl chain ( \(\:\begin{array}{l}{I}_{F{A}_{i}}\end{array}\) ) is calculated as: \(\:\begin{array}{l}{I}_{F{A}_{i}}={I}_{L}\times\:\frac{\text{M}\text{W}\left(F{A}_{i}\right)}{{\text{M}\text{W}}_{L}}\end{array}\) . \(\:{I}_{L}\) is the total measured abundance of the parent lipid, and \(\:{\text{M}\text{W}}_{L}\) is its total molecular weight. The molecular weight of the fatty acyl chain, \(\:\text{M}\text{W}\left(F{A}_{i}\right)\) , is calculated from its chemical structure (number of carbons \(\:{C}_{i}\) and double bonds \(\:{\text{D}\text{B}}_{i}\) ) using the formula: $$\:\begin{array}{l}\text{M}\text{W}\left(F{A}_{i}\right)=12.011\times\:{C}_{i}+1.008\times\:(2{C}_{i}-2{\text{D}\text{B}}_{i})+15.999\times\:2\end{array}$$ . For example, for a triglyceride TG (14:0/22:5/24:0) with a total abundance of \(\:{I}_{L}\) and total molecular weight of \(\:{\text{M}\text{W}}_{L}\) , the attributed abundances of its three fatty acyl chains are calculated as: $$\:\begin{array}{ll}{I}_{{\text{F}\text{A}}_{1}(14:0)}&\:={I}_{L}\times\:\frac{\text{M}\text{W}(14:0)}{{\text{M}\text{W}}_{L}}\\\:{I}_{{\text{F}\text{A}}_{2}(22:5)}&\:={I}_{L}\times\:\frac{\text{M}\text{W}(22:5)}{{\text{M}\text{W}}_{L}}\\\:{I}_{{\text{F}\text{A}}_{3}(24:0)}&\:={I}_{L}\times\:\frac{\text{M}\text{W}(24:0)}{{\text{M}\text{W}}_{L}}.\end{array}$$ After abundance attribution, the attributed abundances of identical fatty acyl moieties were summed independently within each lipid Main Class ( \(\:C\) ) to obtain the total abundance of that fatty acyl chain in the specific class ( \(\:{I}_{\text{t}\text{o}\text{t}\text{a}\text{l}}(\text{F}\text{A},C)\) ): $$\:\begin{array}{l}{I}_{\text{t}\text{o}\text{t}\text{a}\text{l}}(\text{F}\text{A},C)=\sum\:{I}_{F{A}_{i}}\:\:({L}_{j}\in\:C)\end{array}$$ . This process shifts the analytical focus from intact lipid molecules to structurally defined fatty acyl chains within the context of specific lipid classes, laying the foundation for subsequent structure-oriented analysis. Structure-Abundance Correlation Analysis To investigate whether lipid metabolic remodeling during disease follows specific structural rules, we quantified the relationship between changes in lipid abundance and fatty acyl chain structural parameters. Specifically, we calculated the log2 fold change (log 2 FC) for each fatty acyl chain between different clinical comparison groups and used linear regression models to test the correlation of log 2 FC with the number of carbon atoms and double bonds, respectively. Significant positive or negative correlations can indirectly reflect systematic changes in the activity of upstream key metabolic enzymes, such as elongases (ELOVLs) or desaturases (FADS/SCD). Inference and Systemic Assessment of Key Metabolic Enzyme Activities To evaluate changes in key fatty acid metabolic pathways (e.g., elongation and desaturation), we employed a dual analysis strategy. This approach combined the direct inference of individual enzyme activities via product-substrate ratios with a systemic assessment of these pathways across the entire lipidome using structure-abundance correlation analysis. Direct Inference: Enzyme Activity Proxy Analysis via Product-Substrate Ratios To directly infer the activity of key metabolic enzymes, the abundance ratios of specific product-substrate pairs were calculated. These ratios included: SCD-1 (16:1/16:0, 18:1/18:0), FADS1 (20:4/20:3), FADS2 (18:3/18:2), and a general elongase (ELOVL) proxy (18:0/16:0). The statistical significance of the differences in these ratios between clinical groups was assessed using the Wilcoxon rank-sum test. Systemic Assessment: Structure-Abundance Correlation Analysis To systematically test for global trends in fatty acid remodeling, we performed a structure-abundance correlation analysis. This method quantifies the linear relationship between a lipid's structural attributes and its magnitude of abundance change between comparison groups. First, for each lipid molecule (L), the log₂ Fold Change (log 2 FC) was calculated for each group comparison (e.g., AMI vs. HC) using the formula: $$\:\begin{array}{l}{\text{log}}_{2}\text{F}\text{C}L={\text{log}}_{2}\left(\frac{{\stackrel{̄}{I}}_{L,\text{A}\text{M}\text{I}}}{{\stackrel{̄}{I}}_{L,\text{H}\text{C}}}\right)\end{array}$$ where \(\:{\stackrel{̄}{I}}_{L,\text{A}\text{M}\text{I}}\) and \(\:{\stackrel{̄}{I}}_{L,\text{H}\text{C}}\) represent the mean abundance of lipid L in the AMI and HC groups, respectively. Concurrently, two key structural parameters were extracted from each lipid's annotation. \(\:{C}_{L}\) : The total number of carbon atoms in the fatty acyl chains. \(\:{\text{D}\text{B}}_{L}\) : The total number of double bonds in the fatty acyl chains. Next, two independent simple linear regression models were constructed to assess the relationship between these structural parameters and the calculated log 2 FC. Model for Carbon Chain Length: To model the dependency of abundance change on the total number of carbons, the following equation was used: $$\:\begin{array}{l}{{\text{log}}_{2}\text{F}\text{C}}_{L}={\beta\:}_{0}+{\beta\:}_{1}\times\:{C}_{L}+\epsilon\:\end{array}$$ . Model for Double Bond Count: Similarly, to model the dependency on the total number of double bonds, the following equation was used: $$\:\begin{array}{l}{{\text{log}}_{2}\text{F}\text{C}}_{L}={\beta\:}_{0}+{\beta\:}_{1}\times\:{\text{D}\text{B}}_{L}+\epsilon\:\end{array}$$ . For each model, the statistical significance of the relationship was determined by performing a t-test on the slope coefficient (β 1 ), under the null hypothesis that β 1 = 0. The Pearson's correlation coefficient (r) was also calculated to measure the strength and direction of the linear association. The analysis was visualized using scatter plots displaying the data points, the linear regression fit line, and its 95% confidence interval. Statistical Analysis and Visualization Multivariate Pattern Recognition To assess the overall distribution patterns and group differences, we employed Principal Component Analysis (PCA) to visualize the global structure and variance of all samples. To further explore the lipid features that best discriminated between clinical groups, we performed LDA independently on each lipid Category. The LDA models not only evaluated classification accuracy but also identified the key lipid molecules contributing most to group separation. Differential Lipid Analysis We used one-way analysis of variance (ANOVA) followed by Dunnett's post-hoc test to screen for lipids that were significantly altered in the AMI and PRMI groups compared to the HC group. An Adj. P-value < 0.05 was considered statistically significant. The results of the differential analysis were visualized using volcano plots and multiple comparison scatter plots. Dynamic Trend Clustering Analysis To identify lipid modules with co-expression patterns during disease progression (HC → AMI → PRMI), we utilized the Fuzzy C-Means (FCM) clustering algorithm. This analysis, based on the mean standardized expression of each lipid across the three groups, grouped lipids with similar dynamic trends into distinct clusters. The clustering results were visualized using a combination of trend line plots and heatmaps. Structure-Driven Biological Function Interpretation To explore the biological functions associated with differential lipids and specific dynamic trend modules, we employed a structure-driven functional interpretation strategy to establish a direct bridge from molecular structure to biological function. The core of this analysis involved leveraging the unique capabilities of the SwissLipids and LipidMaps databases to identify and map associated protein targets (Uniprot ID) based on the lipids' chemical structures (SMILES format). Subsequently, we used the R package clusterProfiler to perform GO and KEGG pathway enrichment analysis on these protein target sets. The enrichment results were visualized using bar charts or network plots to reveal the core biological processes and signaling pathways associated with lipid metabolic remodeling. Abbreviations Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), Healthy Controls(HC), myocardial infarction (MI), Acute Myocardial Infarction (AMI), Post-PCI Recurrent Myocardial Infarction (PRMI), Quality Control (QC), Glycerolipids (GL), Glycerophospholipids (GP), Triacylglycerols (TG), Phosphatidylcholines (PC) , Phosphatidylethanolamines (PE), Sterol lipids (ST), Sphingolipids (SP), Diacylglycerol (DG), Phosphatidylinositols (PI), Linear Discriminant Analysis (LDA), Sphingomyelins (SM), Ceramides (Cer), Cholesterol esters (ChE), Fuzzy C-Means (FCM), Lysophosphatidylcholines (LysoPC) , Principal Component Analysis (PCA),one-way analysis of variance (ANOVA) . Declarations Funding The work was funded by the Young Elite Scientists Sponsorship Program by CAST (No: 2022QNRC001) and Natural Science Foundation of Liaoning Province (No.2025-MS-250). Data availability statement The R code for the LipiDecipher framework is publicly available on GitHub at: https://github.com/AaronHwang8720/LipiDecipher. Any raw data can be requested by directly contacting the author if the request is reasonable. Ethics approval and consent to participate The collection and use of samples were approved by the Ethics Committee of the First Affiliated Hospital of Dalian Medical University (No. YJ-KS-KY-2022-149), and written informed consent was obtained from all participants. Acknowledgments We thank iPhenome Biotechnology, Inc., for their technical support in terms of untargeted metabolomics. Consent for publication Not applicable. Author contributions P.Y., R.H., S.M.: Methodology, Investigation, Software, Visualization, Writing–original draft. A.H. : Data curation, Software, Editing. Y.Z.: Methodology, Software, Review & editing. X.Y., T.B., D.S. : Data curation, Software, Clinical sample collection. Conflicts of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as potential conflicts of interest. References Huang Y, et al. Lipid profiling identifies modifiable signatures of cardiometabolic risk in children and adolescents with obesity. Nat Med. 2025;31(1):294–305. Hassen CB, et al. Change in lipids before onset of dementia, coronary heart disease, and mortality: A 28-year follow-up Whitehall II prospective cohort study. Alzheimers Dement. 2023;19(12):5518–30. Tavakoli S, Duman E. Uncovering the Potential of Lipid Core Quantification for Predicting Major Adverse Cardiovascular Events. Radiology. 2023;308(2):e231546. Lin LC, et al. Lipid metabolism reprogramming in cardiac fibrosis. Trends Endocrinol Metab. 2024;35(2):164–75. Guo D, et al. Association between the triglyceride-glucose index and impaired cardiovascular fitness in non-diabetic young population. Cardiovasc Diabetol. 2024;23(1):39. Kunutsor SK, Laukkanen JA. Further proof of a paradoxical relationship between high-density lipoprotein levels and adverse cardiovascular outcomes: are there implications for cardiovascular disease prevention? Eur J Prev Cardiol. 2023;30(4):290–2. Patel H, Grant JK, Martin SS. Hypertriglyceridaemia and cardiovascular mortality: insights from a large-scale study. Eur J Prev Cardiol. 2024;31(3):278–9. Abeysinghe R et al. An evidence-based lexical pattern approach for quality assurance of Gene Ontology relations. Brief Bioinform, 2022. 23(3). Kanehisa M, et al. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. 2023;51(D1):D587–92. King-Smith E, et al. Probing the chemical 'reactome' with high-throughput experimentation data. Nat Chem. 2024;16(4):633–43. Hu Y, et al. PANGEA: a new gene set enrichment tool for Drosophila and common research organisms. Nucleic Acids Res. 2023;51(W1):W419–26. Wölk M, Prabutzki P, Fedorova M. Analytical Toolbox to Unlock the Diversity of Oxidized Lipids. Acc Chem Res. 2023;56(7):835–45. Criscuolo A, et al. Analytical and computational workflow for in-depth analysis of oxidized complex lipids in blood plasma. Nat Commun. 2022;13(1):6547. Ni Z, et al. Guiding the choice of informatics software and tools for lipidomics research applications. Nat Methods. 2023;20(2):193–204. Conroy MJ, et al. LIPID MAPS: update to databases and tools for the lipidomics community. Nucleic Acids Res. 2024;52(D1):D1677–82. Aimo L, et al. The SwissLipids knowledgebase for lipid biology. Bioinformatics. 2015;31(17):2860–6. Castro-Alves V, Orešič M, Hyötyläinen T. Lipidomics in nutrition research. Curr Opin Clin Nutr Metab Care. 2022;25(5):311–8. Wu Z, et al. Lipidomics: Mass spectrometric and chemometric analyses of lipids. Adv Drug Deliv Rev. 2020;159:294–307. Buja LM. Lipid abnormalities in myocardial cell injury. Trends Cardiovasc Med. 1991;1(1):40–5. Hadas Y, et al. Altering Sphingolipid Metabolism Attenuates Cell Death and Inflammatory Response After Myocardial Infarction. Circulation. 2020;141(11):916–30. Empinado HM, et al. Diaphragm dysfunction in heart failure is accompanied by increases in neutral sphingomyelinase activity and ceramide content. Eur J Heart Fail. 2014;16(5):519–25. Zhang XJ, et al. Pharmacological inhibition of arachidonate 12-lipoxygenase ameliorates myocardial ischemia-reperfusion injury in multiple species. Cell Metab. 2021;33(10):2059. –2075.e10. McCluskey ER, et al. The arachidonic acid metabolic capacity of canine myocardium is increased during healing of acute myocardial infarction. Circ Res. 1982;51(6):743–50. Bilheimer DW, et al. Fatty acid accumulation and abnormal lipid deposition in peripheral and border zones of experimental myocardial infarcts. J Nucl Med. 1978;19(3):276–83. Hlushchenko I, et al. Linking Variability of Leukocyte Lipid Metabolism to Circulating Lipids, Lipoprotein Composition, and Cardiovascular Risk in the Finnish Adult Population. Arterioscler Thromb Vasc Biol. 2025;45(8):1416–31. Tokutome M, et al. Peroxisome proliferator-activated receptor-gamma targeting nanomedicine promotes cardiac healing after acute myocardial infarction by skewing monocyte/macrophage polarization in preclinical animal models. Cardiovasc Res. 2019;115(2):419–31. Raas Q et al. Metabolic rerouting via SCD1 induction impacts X-linked adrenoleukodystrophy. J Clin Invest, 2021. 131(8). Cavallero S, et al. Exercise mitigates flow recirculation and activates metabolic transducer SCD1 to catalyze vascular protective metabolites. Sci Adv. 2024;10(7):eadj7481. Krajnik A, et al. Phosphoinositide Signaling and Mechanotransduction in Cardiovascular Biology and Disease. Front Cell Dev Biol. 2020;8:595849. Manna P, Jain SK. Phosphatidylinositol-3,4,5-triphosphate and cellular signaling: implications for obesity and diabetes. Cell Physiol Biochem. 2015;35(4):1253–75. Additional Declarations No competing interests reported. Supplementary Files Supplement.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-7238340","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Method Article","associatedPublications":[],"authors":[{"id":500720263,"identity":"648bf16b-1e0e-4043-b33a-cd6f2d243014","order_by":0,"name":"Anliang Huang","email":"","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Anliang","middleName":"","lastName":"Huang","suffix":""},{"id":500720264,"identity":"f74b3ddb-5d5a-4b86-82c6-6988136e2352","order_by":1,"name":"Yunshu Zhang","email":"","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yunshu","middleName":"","lastName":"Zhang","suffix":""},{"id":500720265,"identity":"de1681f2-2e2a-45ce-b780-189e00ad60c2","order_by":2,"name":"Tingting Bai","email":"","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Tingting","middleName":"","lastName":"Bai","suffix":""},{"id":500720266,"identity":"e3104590-0db0-4806-a55a-adeb9204ede8","order_by":3,"name":"Xiaoyang Yuan","email":"","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Xiaoyang","middleName":"","lastName":"Yuan","suffix":""},{"id":500720267,"identity":"5d34984d-64aa-4c36-8293-26a484265200","order_by":4,"name":"Dong Shang","email":"","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Dong","middleName":"","lastName":"Shang","suffix":""},{"id":500720268,"identity":"26393603-ad72-47b3-b86e-43db93b88638","order_by":5,"name":"Shurong Ma","email":"","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Shurong","middleName":"","lastName":"Ma","suffix":""},{"id":500720270,"identity":"04eeac1c-137a-47f4-a03f-4169980d0903","order_by":6,"name":"Rihong Huang","email":"","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Rihong","middleName":"","lastName":"Huang","suffix":""},{"id":500720272,"identity":"e0ce1c4e-4407-44b1-80cc-07df08f7434e","order_by":7,"name":"Peiyuan Yin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA4klEQVRIiWNgGAWjYDACZhBhAGYdkACLHCBeC1sCkVoQgMeAOC0Gx3mPPfhQcDixf3bPx1s32xjk+G4kMH4uwKflMF+64QyDw4kz7pzdbJ3bxmAseSOBWXoGXi08ZtI8QC0NN3K3SQO1JG64kcDGzENIyx+glvk3cp6BtNQTpwVIAg3PYQNpSTAgpEXyMI+5YY9BuvHGG2nG1jnnJAxnnnnYLI1PC9/5M2YPfvyxlp13I/nh7ZwyG3m+48kHP+PTonCAgQ1INcP4oKhhbMCjgYFBvgGspQ6volEwCkbBKBjhAABVuU6O+lqFbAAAAABJRU5ErkJggg==","orcid":"","institution":"The First Affiliated Hospital of Dalian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Peiyuan","middleName":"","lastName":"Yin","suffix":""}],"badges":[],"createdAt":"2025-07-29 03:08:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7238340/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7238340/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":89379332,"identity":"76ae6b7e-2231-4322-812c-cb34a293733d","added_by":"auto","created_at":"2025-08-19 11:37:25","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":819092,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOverall flowchart for LipiDecipher. .\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/608673889b47d5ce08896701.png"},{"id":89378105,"identity":"92adf15a-385c-42ba-87c6-a02d9866f95d","added_by":"auto","created_at":"2025-08-19 11:29:25","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1615813,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGlobal landscape of the lipidome during the progression of myocardial infarction (MI). (\u003c/strong\u003eA) Heatmap of the global lipidomic expression profile during the progression of myocardial infarction. (B) Alluvial plot of lipid composition in category and relative abundance across clinical groups. (C) Hierarchical relationship and abundance changes in lipid classes during MI progression.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/862b5edc7ec45f108fd9a167.png"},{"id":89380312,"identity":"d1f518e1-eda8-4705-a027-77a911c9e88a","added_by":"auto","created_at":"2025-08-19 11:45:25","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":155453,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLinear Discriminant Analysis (LDA) reveals the discriminatory power and key drivers in major lipid categories. \u003c/strong\u003e(A-D) is dedicated to an LDA model for a major lipid Category: (A) Sterol Lipids (ST), (B) Sphingolipids (SP), (C) Glycerophospholipids (GP), and (D) Glycerolipids (GL). Each sub-panel features a central scores plot, with peripheral loading plots surrounding it.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/bf181d77352559b71df97399.png"},{"id":89378108,"identity":"f2a0e7e6-58bb-4c46-8488-de077ec73baf","added_by":"auto","created_at":"2025-08-19 11:29:25","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":312565,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifferential lipid signatures reveal key biological pathways in myocardial infarction recurrence.\u003c/strong\u003e (A) Multi-comparison overview plot. This panel juxtaposes two key comparisons: \"AMI vs. HC\" and \"PRMI vs. AMI\". Point color indicates the direction of change (Up/Down/No change), and size is proportional to the absolute log₂FC. Lipids meeting the significance criteria (Adj. P \u0026lt; 0.05 and |FC| \u0026gt; 1.5) are highlighted and labeled. (B) KEGG pathway enrichment analysis. Based on the predicted protein targets of differential lipids identified between PRMI and AMI groups. The bar plot shows the most significantly enriched pathways. Bar length represents the number of target proteins (Gene Count), and color intensity corresponds to the enrichment significance (-log10(Adj. P)). (C) Gene Ontology (GO) term enrichment analysis. The panel displays the most enriched terms, faceted by the three GO domains: Biological Process (BP), Cellular Component (CC), and Molecular Function (MF). Interpretation of bar length and color scale is consistent with panel (B).\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/cce7c890f7d9b0d5f378c4b2.png"},{"id":89379335,"identity":"cc5dfe9d-f04b-496e-8e80-b8d753f93312","added_by":"auto","created_at":"2025-08-19 11:37:25","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":515921,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDynamic clustering reveals co-regulated lipid modules and their functional pathways during the progression of myocardial infarction (MI). \u003c/strong\u003e(A) Fuzzy C-Means (FCM) clustering of lipid expression profiles and corresponding functional enrichment. Right Panel (Enrichment Dot Plot): the functional enrichment results for the predicted protein targets of lipids within each cluster. Each dot represents a significantly enriched KEGG pathway or Gene Ontology (GO) term. Dot color indicates the ontology source (KEGG or GO domains). (B) Bipartite network of lipids and their assigned expression clusters.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/e981ed8338f6268c7becf51e.png"},{"id":89379334,"identity":"487b2882-adec-4afa-b755-aca665bc5c98","added_by":"auto","created_at":"2025-08-19 11:37:25","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":459237,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLipid deconvolution reveals the structurally-specific remodeling of fatty acid chains in myocardial infarction.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/bbe47a77075b864a5c7c1764.png"},{"id":89378110,"identity":"2cfede34-790c-4e90-974a-b88448fdf362","added_by":"auto","created_at":"2025-08-19 11:29:25","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":515802,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStructurally-resolved lipidomics reveals fatty acid remodeling, altered enzyme activity, and correlations between molecular structure and abundance. \u003c/strong\u003e(A) Fold change in individual fatty acid (FA) chains ccompared to the Healthy Control (HC) group. (B) Comparison of inferred enzyme activities across clinical groups. (C) Correlation between FA abundance changes and structural parameters.\u003c/p\u003e","description":"","filename":"Figure7.png","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/c2b6385178c553af83fc1217.png"},{"id":92248645,"identity":"de5b81cc-7700-4f8f-9f8c-fd1d4e00cb16","added_by":"auto","created_at":"2025-09-26 10:08:53","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6079864,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/57bbd6fd-bbbf-4784-9aff-e9d3542e43bf.pdf"},{"id":89378103,"identity":"1fd8734d-46d9-4577-94a3-792ecf1c8798","added_by":"auto","created_at":"2025-08-19 11:29:25","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":173484,"visible":true,"origin":"","legend":"","description":"","filename":"Supplement.docx","url":"https://assets-eu.researchsquare.com/files/rs-7238340/v1/7069b99f5a7ec4e53955d1d4.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"LipiDecipher: A Structure-Oriented Framework for Mechanistic Interpretation in Clinical Lipidomics","fulltext":[{"header":"Background","content":"\u003cp\u003eLipids are fundamental biological molecules essential for cellular structure, energy homeostasis, and signal transduction. While conventional clinical assays measuring total triglycerides and cholesterol serve as primary predictors for cardiovascular events, their diagnostic specificity and early-prediction sensitivity are limited[\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. The advent of high-throughput mass spectrometry has propelled lipidomics to the forefront of clinical research, offering a comprehensive and specific landscape of circulating lipids that can significantly enhance prognostic accuracy for complex diseases[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. In the post-genomic era, clinical lipidomics, driven by specific clinical questions and the complexity of patient samples, promises to revolutionize early diagnosis, precision medicine, and therapeutic strategies[\u003cspan additionalcitationids=\"CR5 CR6\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eDespite our ability to measure thousands of lipid species, the sheer diversity and structural complexity of the lipidome present formidable analytical and interpretive challenges. Current research is often confined to analyzing individual lipid classes, failing to capture the global, systems-level picture. In other omics disciplines, functional interpretation is greatly facilitated by well-established frameworks like Gene Ontology[\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e] (GO) and pathway enrichment analysis (e.g., Kyoto Encyclopedia of Genes and Genomes[\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e] (KEGG), which effectively translate vast datasets into biological insights. However, the lipidomics field currently lacks a comparably mature analytical paradigm, creating a significant bottleneck.\u003c/p\u003e\u003cp\u003eThis gap stems from a fundamental mismatch between generic analytical logic and the unique nature of lipids. The biological function of a lipid is intrinsically tied to its detailed chemical structure, yet most analytical workflows, particularly those adapted from general metabolomics, fail to account for this. They often overlook the critical fact that subtle structural variations\u0026mdash;such as differences in fatty acyl chain length or the number of double bonds\u0026mdash;can lead to vastly different, or even opposing, biological activities within the same lipid class. This limitation often restricts functional analysis to a superficial level, masking the precise structure-driven mechanisms at play. To unlock the full clinical potential of lipidomics, a new analytical strategy is urgently needed\u0026mdash;one that can systematically link lipid structural features to biological function and enable a more profound, mechanistically-informed interpretation.\u003c/p\u003e\u003cp\u003eTo address this challenge, we developed LipiDecipher, a comprehensive and modular analytical framework designed to navigate the complexities of lipidomics data. The entire workflow, from data acquisition to biological interpretation, is schematically illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. LipiDecipher integrates multi-scale statistical analyses with a structure-centric approach, organized into a dual-pronged analytical strategy to provide deeper biological insights. The first arm of this strategy, Workflow 1: Intact Lipid-Level Analysis, focuses on the intact lipid molecule level. It builds upon traditional statistical analyses by applying a \"lipid-protein-pathway\" mapping to features of interest for functional enrichment. This method links identified lipid structures to their known protein targets and subsequently to established biological pathways, aiming to provide a more precise and mechanistically grounded interpretation than conventional annotation-based methods. The second arm, Workflow 2: Structurally-Resolved Analysis, delves into the sub-molecular level. Its core feature is a lipid deconstruction module that algorithmically infers deep structural properties\u0026mdash;such as fatty acyl composition\u0026mdash;from standard LC-MS/MS data. This capability enables the calculation of enzyme activity proxies and facilitates the direct correlation of these granular features with clinical phenotypes. Through this dual-workflow structure, LipiDecipher is presented as a reproducible and extensible analytical paradigm. To promote its adoption and impact, we provide a detailed case study and open-source R code. By doing so, we aim to empower the research community to discover novel biological mechanisms and robust biomarkers, thereby accelerating the translation of lipidomics findings into clinical practice.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cb\u003eGlobal Landscape of the Lipidome during Myocardial Infarction Progression\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo demonstrate the analytical power of the LipiDecipher framework in a real-world clinical scenario, we performed untargeted lipidomics analysis on samples from three clinical groups: Healthy Controls (HC, n\u0026thinsp;=\u0026thinsp;50), Acute Myocardial Infarction (MI, AMI, n\u0026thinsp;=\u0026thinsp;50), and Post-PCI Recurrent Myocardial Infarction (PRMI, n\u0026thinsp;=\u0026thinsp;35). Before delving into biological differences, we first assessed the stability of our analytical workflow. As shown in the alluvial plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA), all Quality Control (QC) samples exhibited a highly consistent lipid composition pattern, with remarkably similar widths and color proportions of their internal flux bands. This provides strong evidence for the stability and reproducibility of the analytical process, laying a solid foundation for subsequent biological insights.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eFirst, we utilized a global heatmap (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) to reveal overall patterns of change. This plot displays the expression profiles of all lipid molecules, clearly revealing systematic lipid remodeling during disease progression. Compared to the HC group, the AMI and PRMI groups presented multiple clusters of significantly upregulated (brown modules) or downregulated (blue modules) lipids, indicating a global alteration in lipid metabolism. Next, to quantify this change at the compositional level, we employed an alluvial plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) to analyze the compositional shift of major lipid categories. Compared to the HC group (where glycerolipids (GL) constituted\u0026thinsp;~\u0026thinsp;38.9% and glycerophospholipids (GP)\u0026thinsp;~\u0026thinsp;54.1%), the relative abundance of GL increased in the AMI and PRMI groups (to ~\u0026thinsp;40.9% and ~\u0026thinsp;45.5%, respectively), while the relative abundance of GP correspondingly decreased (to 51.6% and 47.2%). This result suggests that the onset and recurrence of MI may be associated with a disrupted balance between lipid energy storage and cell membrane structure.\u003c/p\u003e\u003cp\u003eTo precisely identify the key lipid main classes driving this remodeling and to elucidate their hierarchical relationships, we constructed a composite plot combining a Sankey diagram with lollipop charts (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). The Sankey diagram clearly illustrates the hierarchical composition of the untargeted lipidomics data, showing the distribution from major categories to main classes. The width of the flows is proportional to the number of unique lipid molecules within each branch, clearly revealing the primary constituent classes within each category. For instance, the GL category is predominantly composed of triglycerides (TG), while the GP category mainly comprises phosphatidylcholines (PC) and phosphatidylethanolamines (PE). Corresponding to this structure, the flanking lollipop charts quantify the abundance changes at both the category and main class levels. At the category level, GL and sterol lipids (ST) were upregulated in the AMI and PRMI groups compared to HCs, whereas GP and sphingolipids (SP) showed a downward trend. At the more granular main class level, we observed that the trends of change in the AMI (triangles) and PRMI (circles) groups were highly consistent across most main classes. TGs, cholesterol esters (ChE) as major forms of energy storage, and the signaling molecule diacylglycerol (DG) all showed significant and consistent upregulation (log₂FC range approx. [0.4 to 1]). Conversely, several key phospholipids constituting cell membranes, including PC, PE, and phosphatidylinositols (PI), exhibited significant downregulation (log₂FC range approx. [-0.2 to -0.5]). This strongly implies that the lipid dysregulation pattern established after the acute event persists in recurrent patients. However, subtle differences revealed deeper information: compared to the AMI group, TG, DG, and PI showed more substantial changes in the PRMI group, suggesting that the persistent, high-level dysregulation of these specific lipids may be linked to the mechanisms of MI recurrence.\u003c/p\u003e\u003cp\u003e\u003cb\u003eCategory-Specific LDA for Identifying Efficient Discriminatory Features\u003c/b\u003e\u003c/p\u003e\u003cp\u003eFollowing the global analysis that revealed overall trends, the first major branch of our analytical paradigm\u0026mdash;Intact Lipid-Level Analysis\u0026mdash;aims to deeply investigate the biological significance of intact lipid molecules from three complementary perspectives. First, from a \"static-discriminatory\" viewpoint, we sought to precisely identify the key features that contribute most to distinguishing between the different clinical groups from the complex pool of lipid molecules. To this end, we employed a category-specific Linear Discriminant Analysis (LDA) method (Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). This approach does not perform a single global analysis on all lipids but instead constructs an independent classification model for each major lipid category, thereby enabling a more refined elucidation of the distinct roles of different lipid classes in the pathophysiology of the disease.\u003c/p\u003e\u003cp\u003eFigure \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e centralizes the LDA results for four major lipid categories: ST, SP, GP, and GL. The central scatter plot in each sub-figure shows the projection of samples onto the two main discriminant axes (LD1 and LD2). Through this module, we not only achieved effective group separation but also assigned clear biological meaning to these mathematically constructed discriminant axes. A consistent pattern across all models was that the LD1 axis primarily captured the differences between HC and the disease states (AMI/PRMI), while the LD2 axis mainly distinguished between the two distinct disease stages, acute MI (AMI) and PRMI.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe peripheral bar plots quantify and display the lipid molecules that contribute most to the classification (i.e., those with the highest loadings). This feature is central to LipiDecipher's \"difference attribution\" capability. Through this module, we can clearly identify the key molecules driving different biological comparisons. In the ST category (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA), although there was slightly more overlap between groups, the model (LD1 contribution 95.74%) could still effectively distinguish the HC and PRMI groups, driven primarily by the downregulation of various cholesterol esters (ChE), such as ChE (20:5) and ChE (22:6). In the SP category (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB), the distinction between \"health vs. disease\" (LD1 contribution 88.43%) relied mainly on various long-chain SM and ceramides (Cer). Specifically, the upregulation of molecules like SM (d42:1) and Cer (d41:1) was a significant feature for distinguishing the AMI/PRMI groups. In the GP category (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC), the key drivers for distinguishing \"health vs. disease\" (primarily reflected by the LD1 axis, 85.64% contribution) were mainly the downregulation of various polyunsaturated fatty acid (PUFA)-containing phospholipids, such as PC (38:6), PC (38:4), and PI (34:1). In the GL category (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD), the discriminatory power was almost entirely dominated by various TGand diacylglycerols (DG). For instance, TG (54:3) and DG (36:3) were significantly elevated in the disease groups, becoming core markers for distinguishing them from the HC group.\u003c/p\u003e\u003cp\u003eThe category-specific LDA analysis not only constructed high-precision classifiers for each lipid category (cross-validation showed an average accuracy\u0026thinsp;\u0026gt;\u0026thinsp;95%, Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) but, more importantly, successfully identified key lipid molecules with high discriminatory power from a classification perspective, thus laying a foundation for more rigorous differential quantification and functional exploration.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDifferential Comparison and Functional Mapping: Quantifying Changes at Critical Stages\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAfter LDA provided an overview of key discriminatory features, we next proceeded from a \"pairwise-comparison\" perspective. We employed rigorous statistical tests to precisely quantify the lipids that changed significantly during the critical stages of disease \"onset\" (AMI vs. HC) and \"progression\" (PRMI vs. AMI) (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). However, accurately identifying differential lipids is only the first step. We immediately faced a core challenge in lipidomics analysis: how to link these specific lipid molecules to concrete biological pathways. Traditional enrichment analysis tools are not directly applicable to lipids, which greatly limits our understanding of the mechanisms behind lipid remodeling.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo address this challenge, our LipiDecipher analytical paradigm applies an innovative \"lipid-protein-pathway\" mapping strategy. This strategy serves as a critical bridge connecting lipid structure to biological function. Using the differential lipids most closely associated with MI recurrence identified in the \"PRMI vs. AMI\" comparison as input, we first mapped them to known protein targets (Uniprot ID) via their precise chemical structures (e.g., SMILES format) using the SwissLipids and LipidMaps databases. This step successfully translated changes at the lipid molecule level into information at the protein level.\u003c/p\u003e\u003cp\u003eBased on the resulting set of protein targets, we could then perform standard pathway enrichment analysis using the well-established clusterProfiler tool. The results clearly revealed the biological processes in which these recurrence-associated differential lipids are involved:(1) KEGG pathway enrichment analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB) showed that the protein targets of these lipids were significantly enriched in core lipid pathways such as Glycerophospholipid metabolism, Sphingolipid metabolism, and Arachidonic acid metabolism. (2) GO enrichment analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC) provided more detailed functional annotations. At the Biological Process level, the protein targets were highly associated with phospholipid metabolism and fatty acid metabolism. At the Molecular Function level, they pointed mainly to acyltransferase activity. At the Cellular Component level, they were closely related to the peroxisome and lipid droplet.\u003c/p\u003e\u003cp\u003eIn summary, our proposed analytical workflow not only precisely identified lipid molecules associated with disease recurrence but, more importantly, successfully linked these molecules to specific biological pathways through a structure-driven \"lipid-protein-pathway\" mapping strategy. This revealed that MI recurrence may be associated with inflammation and energy metabolism disorders, fully demonstrating the power of our analytical paradigm in bridging the critical gap from lipid structure to functional insight.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDynamic Trend Clustering: Delineating Co-regulated Functional Modules\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAlthough pairwise differential analysis can pinpoint changes at specific stages, it cannot fully depict the continuous dynamic trajectories of lipid molecules across the entire disease spectrum. Therefore, as the final step from a \"dynamic-panoramic\" perspective, we employed Fuzzy C-Means (FCM) clustering analysis. This was aimed at identifying co-regulated functional modules of lipids from the panoramic view of \"health-acute-recurrence\".\u003c/p\u003e\u003cp\u003eThis analysis partitioned all lipid molecules into five functional modules with distinct dynamic trends (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA, left panels). Each module represents a specific biological response pattern: Cluster 3 (Sustained Upregulation): Lipids were significantly elevated during the AMI stage and remained high or increased further during the PRMI stage. Cluster 5 (Sustained Downregulation): Lipids continuously decreased after the onset of the disease.Cluster 1 (V-shaped - Down then Up): Lipids acutely decreased during the AMI stage but showed a recovery trend in the PRMI stage.Cluster 2 (Acute Upregulation): Lipids sharply increased during the AMI stage but tended to fall back in the PRMI stage.Cluster 4 (Moderate Downregulation): Exhibited a milder downward trend.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe true value of these dynamic modules lies in our ability to assign clear biological functions to each dynamic pattern through our core \"lipid-protein-pathway\" mapping strategy. By performing independent pathway enrichment analysis on each cluster module (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA, right panel), we found that lipids with different dynamic patterns participated in distinctly different biological processes: For example, the protein targets of lipids in the sustainedly upregulated Cluster 3 were significantly enriched in the Phosphatidylinositol signaling system and glycerolipid metabolic process, pointing to signaling and metabolic remodeling that are continuously activated during disease progression. In stark contrast, lipids in the sustainedly downregulated Cluster 5 were associated with lipase activity and the phosphatidylinositol 3-kinase complex, suggesting the inhibition of certain key signaling pathways.\u003c/p\u003e\u003cp\u003eTo further reveal the molecular composition of these functional modules, we constructed a bipartite network graph connecting lipids to their clusters (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). This network clearly displays the lipid main class composition within each dynamic module. For instance, the network visually demonstrates that the sustainedly upregulated Cluster 3 is primarily composed of TG and PC, while the sustainedly downregulated Cluster 5 is rich in Cer and TG. This analysis not only identified co-regulated groups of lipids but also tightly linked specific dynamic patterns, functional pathways, and the chemical classes of lipids.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStructurally-Resolved Analysis: Deconstructing Lipids to Reveal Systematic Metabolic Remodeling\u003c/b\u003e\u003c/p\u003e\u003cp\u003eHaving completed the multi-dimensional analysis of intact lipids, we initiated the second major branch of our analytical paradigm\u0026mdash;Structurally-Resolved Analysis. This was designed to transcend the traditional approach of treating lipids as independent entities and to delve into the sub-molecular structural level. The core of this strategy is Lipid Deconstruction, an algorithm that proportionally attributes the abundance of each parent lipid to its constituent fatty acyl chains. Notably, these fatty acyl chains originate from relatively stable complex lipid pools (e.g., TG, PC, SM), and their abundance changes are more indicative of long-term metabolic adaptation and remodeling rather than the transient fluctuations of free fatty acids in the blood.\u003c/p\u003e\u003cp\u003eAfter applying the deconstruction algorithm, we first constructed a global heatmap (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e) to visually present the overall abundance patterns of fatty acyl chains across all samples. In this plot, each row represents a unique fatty acyl chain, and each column represents a biological sample, with color intensity reflecting its standardized abundance. Through the side row annotations, readers can quickly identify the structural features (e.g., saturation, chain length) of each fatty acyl chain and its primary lipid class of origin. This map provides a macroscopic, structured data landscape for subsequent quantitative analysis and offers initial hints of systematic differences between the clinical groups.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eTo examine these differences in greater detail, we further calculated the fold change of each fatty acyl chain relative to the HC group within each major lipid class (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA). This plot revealed complex, class-specific patterns, highlighting the necessity of analyzing fatty acyl chains on different lipid backbones independently. (1) Heterogeneous changes in the GP family: We observed significant heterogeneity within the GP family. In contrast to the downward or stable trend of most fatty acyl chains in PC and PE, those in lysophosphatidylcholines (LysoPC) showed a moderate and general upregulation. Particularly noteworthy were the fatty acids associated with PI. Although their changes were relatively modest, they were consistently downregulated in the disease groups, with a significantly greater decrease in the recurrent group (PRMI, circles) than in the acute group (AMI, triangles), presenting a \"dose-effect\" pattern possibly related to disease severity or progression. (2) Unique patterns within TG: Although TGs were generally upregulated, this plot also revealed some unique molecular changes. For example, the saturated fatty acid TG (10:0) showed a sharp, nearly 4-fold increase in the AMI group but returned to near-control levels in the PRMI group. This suggests that TG (10:0) might be a transiently changing molecule closely associated with the acute-phase stress response, rather than a persistent marker. (3) Trends in other classes: Meanwhile, fatty acyl chains in Cer, ChE, and DG also exhibited a general upward trend similar to TGs, especially in the polyunsaturated fraction. The changes in SM were more complex and showed some saturation specificity.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eCollectively, this fine-grained, class-specific analysis not only confirms that fatty acid metabolism on different lipid backbones is differentially regulated but also successfully identifies key fatty acyl chains with unique dynamic behaviors (such as the progressive decrease of PI and the acute increase of TG (10:0), providing more precise targets for subsequent mechanistic investigation and biomarker discovery.\u003c/p\u003e\u003cp\u003eTo systematically test the overall trends of fatty acid elongation and desaturation on a global scale, we performed a structure-abundance correlation analysis. This analysis quantifies the linear relationship between the structural parameters of fatty acyl chains (carbon number and double bond count) and their abundance changes (log₂FC). The results showed that in the \"AMI vs. HC\" and \"PRMI vs. HC\" comparisons, the carbon chain length of fatty acids was positively correlated with their log₂FC values (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB). This finding is consistent with the activation of the overall fatty acid elongation process. The number of double bonds, however, did not show a clear correlation, suggesting that the desaturation process may be less strongly associated with the disease state.\u003c/p\u003e\u003cp\u003eHowever, correlation analysis can only reflect overall trends and cannot directly point to the activity changes of specific enzymes. Therefore, we finally calculated specific product-to-substrate ratios as proxies for enzyme activity to directly infer changes in key metabolic enzymes. The results (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC, D) showed that compared to the HC group, the activity proxy for stearoyl-CoA desaturase (SCD-18, 18:1/18:0) was significantly elevated in both the AMI and PRMI groups (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Although the activity proxies for elongase (ELOVL) and other desaturases (SCD-16, FADS1, FADS2) did not show statistically significant differences between groups, the clear change in SCD-18 activity confirms that even within the stable complex lipid pool, the activation of the desaturation pathway is a detectable key event in the pathophysiology of MI.\u003c/p\u003e\u003cp\u003eIn summary, by integrating lipid deconstruction, class-specific quantification, structure-abundance correlation, and enzyme activity proxy analysis, our analytical paradigm successfully revealed the systematic remodeling of fatty acid metabolism in myocardial infarction from a structural level. This provides deeper, structure-driven insights for understanding the mechanisms of the disease. With this, we have completed a comprehensive exploration from the global landscape to intact molecules and down to sub-molecular structures, laying a solid foundation for the subsequent discussion.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we utilized the LipiDecipher analytical framework to conduct a systematic, multi-dimensional analysis of lipidomics data related to myocardial infarction. This application demonstrated the framework's robust capabilities in processing complex lipidomics data, identifying potential biomarkers, and elucidating disease-associated biological mechanisms. More importantly, it offers a practical solution for the paradigm shift in lipidomics research, moving from molecular identification to functional interpretation.\u003c/p\u003e\u003cp\u003e\u003cb\u003eLipiDecipher: A Synergistic Strategy Integrating Multi-Source Information and Deconstructing Lipid Structure\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAccurate lipid identification and annotation serve as the critical link between raw data and biological meaning in lipidomics research. However, the inherent limitations of single databases and the structural complexity of lipids often lead to incomplete or inaccurate annotations[\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. LipiDecipher addresses these challenges through two core designs.\u003c/p\u003e\u003cp\u003eFirstly, by integrating multiple authoritative databases[\u003cspan additionalcitationids=\"CR15\" citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e] (e.g., Lipid Maps, SwissLipids), LipiDecipher builds a more comprehensive and robust annotation system. This integration is not merely a data aggregation but a strategy of cross-validation and information complementarity. It significantly enhances annotation accuracy and coverage, effectively mitigating the \"information silo\" problem caused by database discrepancies and providing a high-quality, structured data foundation for subsequent lipid-protein-pathway mapping.\u003c/p\u003e\u003cp\u003eSecondly, a hallmark feature of LipiDecipher is its lipid structure deconstruction capability. Whereas traditional analysis often stops at identifying intact molecules like TG (54:3), our framework delves deeper by deconstructing them into their constituent fatty acyl chains. As demonstrated in our analysis of Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e, this ability to move from macroscopic molecules to microscopic structures allows for a systematic evaluation of how fatty acids with different chain lengths and saturation levels change during disease progression within a class-specific context. This unique analytical dimension enables LipiDecipher to uncover the deeper, structure-driven mechanisms hidden beneath the changes in intact lipid molecules.\u003c/p\u003e\u003cp\u003e\u003cb\u003eSystematic Lipid Remodeling: From Global Landscape to Dynamic Modules\u003c/b\u003e\u003c/p\u003e\u003cp\u003eOur investigation began at a global level, revealing a profound and systematic lipid remodeling during MI and its recurrence. A salient feature was the fundamental shift in lipid composition from healthy controls to AMI and PRMI patients: a progressive increase in the relative abundance of glycerolipids (GL) for energy storage, with a concomitant decrease in glycerophospholipids (GP), the structural backbone of cell membranes (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). This finding established a key theme for our study\u0026mdash;that the pathophysiology of MI involves a dysregulation of the balance between energy metabolism and cell membrane homeostasis.\u003c/p\u003e\u003cp\u003eTo comprehensively capture the key molecules driving this shift, we employed a multi-dimensional screening strategy: Static Discrimination (LDA, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), from a classification perspective, efficiently identified the most discriminating lipids. For instance, the downregulation of various PUFA-containing phospholipids (e.g., PC (38:6), PI (34:1)) and the upregulation of long-chain sphingolipids (e.g., SM(d42:1), Cer(d41:1)) were core drivers distinguishing the 'healthy vs. disease' states. Dynamic Clustering (FCM, Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e), from a panoramic disease perspective, revealed co-regulated lipid modules. For example, the 'Sustained Upregulation' Cluster 3, primarily composed of triglycerides (TG) and phosphatidylcholines (PC), was functionally enriched in the \"Phosphatidylinositol signaling system,\" suggesting its continuous activation. Conversely, the 'Sustained Downregulation' Cluster 5, rich in ceramides (Cer), was associated with the inhibition of \"lipase activity.\"\u003c/p\u003e\u003cp\u003eThrough this progressive analysis, from global composition to static discrimination and dynamic patterns, we cross-validated the central role of key lipid classes such as TGs, PCs, and Cers in the pathology of MI, laying a solid foundation for subsequent mechanistic exploration.\u003c/p\u003e\u003cp\u003e\u003cb\u003e\"Lipid-Protein-Pathway\" Mapping: Overcoming the Hurdle of Functional Enrichment in Lipidomics\u003c/b\u003e\u003c/p\u003e\u003cp\u003eOne of the major bottlenecks in current lipidomics research is linking identified differential lipids to specific biological pathways[\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. To address this challenge, we proposed an innovative \"lipid-protein-pathway\" mapping strategy. By predicting the protein targets of differential lipids using databases like SwissLipids[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e], we successfully shifted the analytical dimension from lipids to proteins, thereby enabling the use of well-established pathway enrichment tools.\u003c/p\u003e\u003cp\u003eIn our study (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), we applied this strategy to focus on the differential lipids most closely associated with MI recurrence (PRMI vs. AMI). Their protein targets were successfully mapped to core pathways, including Glycerophospholipid metabolism, Sphingolipid metabolism, and Arachidonic acid metabolism. These findings are highly consistent with existing knowledge: dysregulation of glycerophospholipid metabolism is directly linked to cardiomyocyte injury and dysfunction[\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], the role of sphingolipids (especially ceramides) in apoptosis and inflammation makes them a focal point in MI research[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], and arachidonic acid metabolites, as key inflammatory mediators, play an undisputed role in MI pathophysiology[\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eFurthermore, GO enrichment analysis pointed to cellular components such as lipid droplets and peroxisomes, forming a logical loop with our observations of altered energy metabolism (TG accumulation) and potential oxidative stress at the global level[\u003cspan additionalcitationids=\"CR25\" citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. These results not only validate the effectiveness of our mapping strategy but also provide a new, systems-biology perspective for understanding the mechanisms of MI recurrence.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStructural Deconstruction: Revealing Deeper Metabolic Mechanisms from Fatty Acyl Chain Remodeling\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThrough LipiDecipher's unique deconstruction analysis, we were able to transcend the analysis of intact lipids and directly investigate the association between fatty acyl chain structure and disease state. Importantly, our analysis is based on stable, complex lipid pools, making our findings more reflective of long-term metabolic remodeling Our analysis revealed complex and biologically meaningful patterns. First, the structure-abundance correlation analysis in Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eB showed that the carbon chain length of fatty acids was positively correlated with their abundance changes, suggesting that the fatty acid elongation process is systematically activated during the disease. However, the number of double bonds showed no clear correlation, indicating that the desaturation process might be more complex.\u003c/p\u003e\u003cp\u003eTo investigate this complexity, we calculated proxies for enzyme activity (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eC). The results showed that the activity of stearoyl-CoA desaturase (SCD-18) was significantly elevated, which aligns with existing literature on the role of SCD1 in cardiovascular protection and anti-inflammation[\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. In contrast, other desaturases (FADS1/2) showed no significant changes. This suggests that, against a background of unclear overall desaturation trends, the SCD-18 pathway is specifically activated and may be a key metabolic node driving disease progression.\u003c/p\u003e\u003cp\u003eNotably, the class-specific deconstruction analysis (Fig.\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003eA) revealed fine-grained regulatory mechanisms unattainable by traditional methods: Progressive Deterioration Pattern: Our deconstruction analysis found that fatty acyl chains associated with PI exhibited a \"dose-effect\" pattern, with a significantly greater decrease in the PRMI group than in the AMI group. This pattern strongly suggests a continuous deterioration of cell membrane signaling and repair capacity as the disease progresses. The biological basis for this inference lies in the central role of PI and its phosphorylated derivatives as key nodes in cellular signaling networks. For instance, the critical signaling molecule PtdIns(3,4,5) P₃ regulates cell survival, proliferation, and response to mechanical stress by activating downstream proteins\u0026mdash;processes crucial for cardiomyocyte repair after ischemic injury[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. Thus, the unique insight from LipiDecipher is that it not only confirms the central role of the PI signaling pathway in MI pathology but also reveals for the first time that this dysfunction may be linked to the progressive depletion of the PI molecular pool itself (specifically its fatty acyl chains), offering a deeper, structure-driven perspective on the mechanisms of disease exacerbation.\u003c/p\u003e\u003cp\u003eAcute Stress Pattern: The short-chain saturated fatty acid TG (10:0) showed a sharp, nearly 4-fold increase exclusively in the AMI group, returning to baseline in the PRMI group. This strongly suggests that it is a transient marker closely associated with the acute-phase stress response, rather than a persistent marker of recurrence. These findings fully demonstrate the unique value of deconstruction analysis in revealing fine-grained, structure-driven regulatory mechanisms.\u003c/p\u003e\u003cp\u003e\u003cb\u003eSignificance and Limitations\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThe core contribution of this study is that we not only revealed a series of profound lipid metabolic disturbances during MI and its recurrence but, more importantly, we proposed and validated a systematic, structure-oriented lipidomics analysis paradigm named LipiDecipher. Conceptually, it advances lipidomics research from \"discovering differential molecules\" to \"understanding structure-function relationships.\" Technically, through its \"multi-dimensional feature screening\" and \"lipid-protein-pathway\" mapping, it provides a viable approach to address the long-standing challenges of biomarker reliability and functional annotation in the field.\u003c/p\u003e\u003cp\u003eHowever, it is important to acknowledge several limitations. First, as previously mentioned, all protein targets and pathways predicted bioinformatically require subsequent experimental validation. Second, this study's cross-sectional design reveals association, not causation; future longitudinal cohort studies are warranted to track the dynamic changes of lipids and validate their prognostic value. Finally, our findings are based on a specific clinical cohort, and their generalizability needs to be tested in larger, more diverse populations.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eBy applying the LipiDecipher analytical framework, this study systematically delineated the multi-level lipid remodeling landscape during the onset and recurrence of myocardial infarction and successfully linked specific lipid structural changes to key metabolic pathways. Our proposed framework demonstrates its potential in handling complex lipidomics data, identifying reliable biomarkers, and revealing disease-related biological mechanisms. We believe that LipiDecipher, as an open-source and extensible tool, will help advance clinical lipidomics research and accelerate the translation of lipidomics findings into novel strategies for improving cardiovascular disease diagnosis and treatment. The R code for the LipiDecipher framework is publicly available on GitHub at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/AaronHwang8720/LipiDecipher\u003c/span\u003e\u003cspan address=\"https://github.com/AaronHwang8720/LipiDecipher\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cb\u003eClinical Cohort and Sample Information\u003c/b\u003e\u003c/p\u003e\u003cp\u003eThis study enrolled a clinical cohort comprising three groups of subjects: a healthy control group (HC, n\u0026thinsp;=\u0026thinsp;50), an acute myocardial infarction group (AMI, n\u0026thinsp;=\u0026thinsp;50), and a group of patients with post-percutaneous coronary intervention (PCI) recurrent myocardial infarction (PRMI, n\u0026thinsp;=\u0026thinsp;35). Detailed clinical baseline characteristics of the cohort are provided in Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eUHPLC-MS/MS Lipidomics Analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eSample Preparation and Lipid Extraction\u003c/b\u003e\u003c/p\u003e\u003cp\u003eLipid extraction was performed using a methyl tert-butyl ether (MTBE) liquid-liquid extraction method. The core extraction procedure was as follows: 120 \u0026micro;L of methanol containing internal standards was added to 20 \u0026micro;L of serum, and the mixture was vortexed (1000 rpm, 5 min). Subsequently, 360 \u0026micro;L of MTBE and 100 \u0026micro;L of ultrapure water were added to induce phase separation, followed by another vortexing step (1000 rpm, 10 min) and incubation at room temperature for 10 minutes. After centrifugation at 13,000 g for 15 min at 4\u0026deg;C, the upper organic phase was collected and lyophilized. Prior to analysis, the lipid residue was reconstituted in 80 \u0026micro;L of acetonitrile-isopropanol (1:1, v/v). Quality control (QC) samples were prepared by pooling equal aliquots from all biological samples and were processed alongside them.\u003c/p\u003e\u003cp\u003e\u003cb\u003eUntargeted Lipidomics Analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eUntargeted analysis was conducted on a UPLC-HRMS system, consisting of a Thermo Scientific Ultimate\u0026trade; 3000 UPLC system coupled to a Thermo Scientific Q Exactive\u0026trade; Quadrupole-Orbitrap high-resolution mass spectrometer (Thermo Scientific, USA). Chromatographic separation was achieved using an Accucore C30 core-shell column. The mobile phase consisted of (A) 60% acetonitrile in water and (B) 90% isopropanol/10% acetonitrile, both containing ammonium formate and 0.1% formic acid. The gradient elution program was as follows: initial 10% B, ramped linearly to 50% B in 5 min, then further ramped to 100% B over 23 min.\u003c/p\u003e\u003cp\u003eThe mass spectrometer was equipped with a heated electrospray ionization (H-ESI) source, and data were acquired in both positive and negative ionization modes. Ionization parameters were set as follows: sheath gas flow rate, 45 arb; auxiliary gas flow rate, 10 arb; capillary temperature, 320\u0026deg;C; heater temperature, 355\u0026deg;C; S-Lens RF level, 55. Data were collected in full scan mode.\u003c/p\u003e\u003cp\u003e\u003cb\u003eLipidomics Data Preprocessing\u003c/b\u003e\u003c/p\u003e\u003cp\u003eRaw mass spectrometry data underwent a series of standardized preprocessing steps. In brief, the data were first subjected to quality control (QC), where lipid features were filtered out if they had a missing value rate\u0026thinsp;\u0026gt;\u0026thinsp;50% in QC samples or \u0026gt;\u0026thinsp;20% in biological samples. Unstable features with a coefficient of variation (CV)\u0026thinsp;\u0026gt;\u0026thinsp;30% in QC samples were also removed. Subsequently, remaining missing values were imputed using the k-Nearest Neighbors (kNN) algorithm. Finally, all lipids were annotated, assigned unique lipid IDs, and classified into different Categories and Main Classes based on their chemical structures.\u003c/p\u003e\u003cp\u003e\u003cb\u003eLipiDecipher: A Structure-Oriented Lipidomics Analysis Framework\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo extract structure-function information from complex lipidomics data, we developed and applied the LipiDecipher analysis framework. This framework includes the following core modules:\u003c/p\u003e\u003cp\u003e\u003cb\u003eLipid Deconstruction and Abundance Attribution\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTraditional lipid analysis treats each lipid molecule as an independent entity. To overcome this limitation, we employed a \"deconstruction\" strategy. This algorithmic approach proportionally attributes the instrumental response (abundance) of each parent lipid to its constituent fatty acyl chains based on their molecular weight relative to the total molecular weight of the parent lipid.\u003c/p\u003e\u003cp\u003eFirst, for a parent lipid (\u003cem\u003eL\u003c/em\u003e), the attributed abundance of a single fatty acyl chain (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\begin{array}{l}{I}_{F{A}_{i}}\\end{array}\\)\u003c/span\u003e\u003c/span\u003e) is calculated as:\u003c/p\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\begin{array}{l}{I}_{F{A}_{i}}={I}_{L}\\times\\:\\frac{\\text{M}\\text{W}\\left(F{A}_{i}\\right)}{{\\text{M}\\text{W}}_{L}}\\end{array}\\)\u003c/span\u003e\u003c/span\u003e.\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{I}_{L}\\)\u003c/span\u003e\u003c/span\u003e is the total measured abundance of the parent lipid, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{M}\\text{W}}_{L}\\)\u003c/span\u003e\u003c/span\u003e is its total molecular weight. The molecular weight of the fatty acyl chain, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{M}\\text{W}\\left(F{A}_{i}\\right)\\)\u003c/span\u003e\u003c/span\u003e, is calculated from its chemical structure (number of carbons \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{C}_{i}\\)\u003c/span\u003e\u003c/span\u003e and double bonds \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{D}\\text{B}}_{i}\\)\u003c/span\u003e\u003c/span\u003e) using the formula:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{l}\\text{M}\\text{W}\\left(F{A}_{i}\\right)=12.011\\times\\:{C}_{i}+1.008\\times\\:(2{C}_{i}-2{\\text{D}\\text{B}}_{i})+15.999\\times\\:2\\end{array}$$\u003c/div\u003e\u003c/div\u003e.\u003c/p\u003e\u003cp\u003eFor example, for a triglyceride TG (14:0/22:5/24:0) with a total abundance of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{I}_{L}\\)\u003c/span\u003e\u003c/span\u003e and total molecular weight of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{M}\\text{W}}_{L}\\)\u003c/span\u003e\u003c/span\u003e, the attributed abundances of its three fatty acyl chains are calculated as:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{ll}{I}_{{\\text{F}\\text{A}}_{1}(14:0)}\u0026amp;\\:={I}_{L}\\times\\:\\frac{\\text{M}\\text{W}(14:0)}{{\\text{M}\\text{W}}_{L}}\\\\\\:{I}_{{\\text{F}\\text{A}}_{2}(22:5)}\u0026amp;\\:={I}_{L}\\times\\:\\frac{\\text{M}\\text{W}(22:5)}{{\\text{M}\\text{W}}_{L}}\\\\\\:{I}_{{\\text{F}\\text{A}}_{3}(24:0)}\u0026amp;\\:={I}_{L}\\times\\:\\frac{\\text{M}\\text{W}(24:0)}{{\\text{M}\\text{W}}_{L}}.\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eAfter abundance attribution, the attributed abundances of identical fatty acyl moieties were summed independently within each lipid Main Class (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:C\\)\u003c/span\u003e\u003c/span\u003e) to obtain the total abundance of that fatty acyl chain in the specific class (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{I}_{\\text{t}\\text{o}\\text{t}\\text{a}\\text{l}}(\\text{F}\\text{A},C)\\)\u003c/span\u003e\u003c/span\u003e):\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{l}{I}_{\\text{t}\\text{o}\\text{t}\\text{a}\\text{l}}(\\text{F}\\text{A},C)=\\sum\\:{I}_{F{A}_{i}}\\:\\:({L}_{j}\\in\\:C)\\end{array}$$\u003c/div\u003e\u003c/div\u003e.\u003c/p\u003e\u003cp\u003eThis process shifts the analytical focus from intact lipid molecules to structurally defined fatty acyl chains within the context of specific lipid classes, laying the foundation for subsequent structure-oriented analysis.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStructure-Abundance Correlation Analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo investigate whether lipid metabolic remodeling during disease follows specific structural rules, we quantified the relationship between changes in lipid abundance and fatty acyl chain structural parameters. Specifically, we calculated the log2 fold change (log\u003csub\u003e2\u003c/sub\u003eFC) for each fatty acyl chain between different clinical comparison groups and used linear regression models to test the correlation of log\u003csub\u003e2\u003c/sub\u003eFC with the number of carbon atoms and double bonds, respectively. Significant positive or negative correlations can indirectly reflect systematic changes in the activity of upstream key metabolic enzymes, such as elongases (ELOVLs) or desaturases (FADS/SCD).\u003c/p\u003e\u003cp\u003e\u003cb\u003eInference and Systemic Assessment of Key Metabolic Enzyme Activities\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo evaluate changes in key fatty acid metabolic pathways (e.g., elongation and desaturation), we employed a dual analysis strategy. This approach combined the direct inference of individual enzyme activities via product-substrate ratios with a systemic assessment of these pathways across the entire lipidome using structure-abundance correlation analysis.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDirect Inference: Enzyme Activity Proxy Analysis via Product-Substrate Ratios\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo directly infer the activity of key metabolic enzymes, the abundance ratios of specific product-substrate pairs were calculated. These ratios included: SCD-1 (16:1/16:0, 18:1/18:0), FADS1 (20:4/20:3), FADS2 (18:3/18:2), and a general elongase (ELOVL) proxy (18:0/16:0). The statistical significance of the differences in these ratios between clinical groups was assessed using the Wilcoxon rank-sum test.\u003c/p\u003e\u003cp\u003e\u003cb\u003eSystemic Assessment: Structure-Abundance Correlation Analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo systematically test for global trends in fatty acid remodeling, we performed a structure-abundance correlation analysis. This method quantifies the linear relationship between a lipid's structural attributes and its magnitude of abundance change between comparison groups.\u003c/p\u003e\u003cp\u003eFirst, for each lipid molecule (L), the log₂ Fold Change (log\u003csub\u003e2\u003c/sub\u003eFC) was calculated for each group comparison (e.g., AMI vs. HC) using the formula:\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{l}{\\text{log}}_{2}\\text{F}\\text{C}L={\\text{log}}_{2}\\left(\\frac{{\\stackrel{̄}{I}}_{L,\\text{A}\\text{M}\\text{I}}}{{\\stackrel{̄}{I}}_{L,\\text{H}\\text{C}}}\\right)\\end{array}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\stackrel{̄}{I}}_{L,\\text{A}\\text{M}\\text{I}}\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\stackrel{̄}{I}}_{L,\\text{H}\\text{C}}\\)\u003c/span\u003e\u003c/span\u003e represent the mean abundance of lipid L in the AMI and HC groups, respectively. Concurrently, two key structural parameters were extracted from each lipid's annotation.\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{C}_{L}\\)\u003c/span\u003e\u003c/span\u003e: The total number of carbon atoms in the fatty acyl chains.\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\text{D}\\text{B}}_{L}\\)\u003c/span\u003e\u003c/span\u003e: The total number of double bonds in the fatty acyl chains.\u003c/p\u003e\u003cp\u003eNext, two independent simple linear regression models were constructed to assess the relationship between these structural parameters and the calculated log\u003csub\u003e2\u003c/sub\u003eFC.\u003c/p\u003e\u003cp\u003eModel for Carbon Chain Length: To model the dependency of abundance change on the total number of carbons, the following equation was used:\u003cdiv id=\"Eque\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Eque\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{l}{{\\text{log}}_{2}\\text{F}\\text{C}}_{L}={\\beta\\:}_{0}+{\\beta\\:}_{1}\\times\\:{C}_{L}+\\epsilon\\:\\end{array}$$\u003c/div\u003e\u003c/div\u003e.\u003c/p\u003e\u003cp\u003eModel for Double Bond Count: Similarly, to model the dependency on the total number of double bonds, the following equation was used:\u003cdiv id=\"Equf\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equf\" name=\"EquationSource\"\u003e\n$$\\:\\begin{array}{l}{{\\text{log}}_{2}\\text{F}\\text{C}}_{L}={\\beta\\:}_{0}+{\\beta\\:}_{1}\\times\\:{\\text{D}\\text{B}}_{L}+\\epsilon\\:\\end{array}$$\u003c/div\u003e\u003c/div\u003e.\u003c/p\u003e\u003cp\u003eFor each model, the statistical significance of the relationship was determined by performing a t-test on the slope coefficient (β\u003csub\u003e1\u003c/sub\u003e), under the null hypothesis that β\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;0. The Pearson's correlation coefficient (r) was also calculated to measure the strength and direction of the linear association. The analysis was visualized using scatter plots displaying the data points, the linear regression fit line, and its 95% confidence interval.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStatistical Analysis and Visualization\u003c/b\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eMultivariate Pattern Recognition\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo assess the overall distribution patterns and group differences, we employed Principal Component Analysis (PCA) to visualize the global structure and variance of all samples. To further explore the lipid features that best discriminated between clinical groups, we performed LDA independently on each lipid Category. The LDA models not only evaluated classification accuracy but also identified the key lipid molecules contributing most to group separation.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDifferential Lipid Analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eWe used one-way analysis of variance (ANOVA) followed by Dunnett's post-hoc test to screen for lipids that were significantly altered in the AMI and PRMI groups compared to the HC group. An Adj. P-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 was considered statistically significant. The results of the differential analysis were visualized using volcano plots and multiple comparison scatter plots.\u003c/p\u003e\u003cp\u003e\u003cb\u003eDynamic Trend Clustering Analysis\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo identify lipid modules with co-expression patterns during disease progression (HC \u0026rarr; AMI \u0026rarr; PRMI), we utilized the Fuzzy C-Means (FCM) clustering algorithm. This analysis, based on the mean standardized expression of each lipid across the three groups, grouped lipids with similar dynamic trends into distinct clusters. The clustering results were visualized using a combination of trend line plots and heatmaps.\u003c/p\u003e\u003cp\u003e\u003cb\u003eStructure-Driven Biological Function Interpretation\u003c/b\u003e\u003c/p\u003e\u003cp\u003eTo explore the biological functions associated with differential lipids and specific dynamic trend modules, we employed a structure-driven functional interpretation strategy to establish a direct bridge from molecular structure to biological function. The core of this analysis involved leveraging the unique capabilities of the SwissLipids and LipidMaps databases to identify and map associated protein targets (Uniprot ID) based on the lipids' chemical structures (SMILES format). Subsequently, we used the R package clusterProfiler to perform GO and KEGG pathway enrichment analysis on these protein target sets. The enrichment results were visualized using bar charts or network plots to reveal the core biological processes and signaling pathways associated with lipid metabolic remodeling.\u003c/p\u003e"},{"header":"Abbreviations","content":"\u003cp\u003eGene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), Healthy Controls(HC), myocardial infarction (MI), Acute Myocardial Infarction (AMI), Post-PCI Recurrent Myocardial Infarction (PRMI), Quality Control (QC), Glycerolipids (GL), Glycerophospholipids (GP), Triacylglycerols (TG), Phosphatidylcholines (PC) , Phosphatidylethanolamines (PE), Sterol lipids (ST), Sphingolipids (SP), Diacylglycerol (DG), Phosphatidylinositols (PI), Linear Discriminant Analysis (LDA), Sphingomyelins (SM), Ceramides (Cer), Cholesterol esters (ChE), Fuzzy C-Means (FCM), Lysophosphatidylcholines (LysoPC) , Principal Component Analysis (PCA),one-way analysis of variance (ANOVA) .\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe work was funded by the Young Elite Scientists Sponsorship Program by CAST (No: 2022QNRC001) and Natural Science Foundation of Liaoning Province (No.2025-MS-250).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability statement\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe R code for the LipiDecipher framework is publicly available on GitHub at: https://github.com/AaronHwang8720/LipiDecipher. Any raw data can be requested by directly contacting the author if the request is reasonable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe collection and use of samples were approved by the Ethics Committee of the First Affiliated Hospital of Dalian Medical University (No. YJ-KS-KY-2022-149), and written informed consent was obtained from all participants.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank iPhenome Biotechnology, Inc., for their technical support in terms of untargeted metabolomics.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConsent for publication\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eP.Y.,\u003c/strong\u003e\u003cstrong\u003e\u0026nbsp;R.H., \u003cstrong\u003eS.M.:\u003c/strong\u003e\u003c/strong\u003e Methodology, Investigation, Software, Visualization, Writing\u0026ndash;original draft. \u003cstrong\u003eA.H.\u003cstrong\u003e:\u003c/strong\u003e\u003c/strong\u003e Data curation, Software, Editing.\u003cstrong\u003e\u0026nbsp;Y.Z.:\u0026nbsp;\u003c/strong\u003eMethodology, Software, Review \u0026amp; editing. \u003cstrong\u003eX.Y., T.B., D.S.\u003cstrong\u003e:\u003c/strong\u003e\u003c/strong\u003e Data curation, Software, Clinical sample collection.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003einterest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as potential conflicts of interest.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eHuang Y, et al. Lipid profiling identifies modifiable signatures of cardiometabolic risk in children and adolescents with obesity. Nat Med. 2025;31(1):294\u0026ndash;305.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHassen CB, et al. Change in lipids before onset of dementia, coronary heart disease, and mortality: A 28-year follow-up Whitehall II prospective cohort study. Alzheimers Dement. 2023;19(12):5518\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTavakoli S, Duman E. Uncovering the Potential of Lipid Core Quantification for Predicting Major Adverse Cardiovascular Events. Radiology. 2023;308(2):e231546.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLin LC, et al. Lipid metabolism reprogramming in cardiac fibrosis. Trends Endocrinol Metab. 2024;35(2):164\u0026ndash;75.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGuo D, et al. Association between the triglyceride-glucose index and impaired cardiovascular fitness in non-diabetic young population. Cardiovasc Diabetol. 2024;23(1):39.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKunutsor SK, Laukkanen JA. Further proof of a paradoxical relationship between high-density lipoprotein levels and adverse cardiovascular outcomes: are there implications for cardiovascular disease prevention? Eur J Prev Cardiol. 2023;30(4):290\u0026ndash;2.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePatel H, Grant JK, Martin SS. Hypertriglyceridaemia and cardiovascular mortality: insights from a large-scale study. Eur J Prev Cardiol. 2024;31(3):278\u0026ndash;9.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAbeysinghe R et al. An evidence-based lexical pattern approach for quality assurance of Gene Ontology relations. Brief Bioinform, 2022. 23(3).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKanehisa M, et al. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. 2023;51(D1):D587\u0026ndash;92.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKing-Smith E, et al. Probing the chemical 'reactome' with high-throughput experimentation data. Nat Chem. 2024;16(4):633\u0026ndash;43.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHu Y, et al. PANGEA: a new gene set enrichment tool for Drosophila and common research organisms. Nucleic Acids Res. 2023;51(W1):W419\u0026ndash;26.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eW\u0026ouml;lk M, Prabutzki P, Fedorova M. Analytical Toolbox to Unlock the Diversity of Oxidized Lipids. Acc Chem Res. 2023;56(7):835\u0026ndash;45.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCriscuolo A, et al. Analytical and computational workflow for in-depth analysis of oxidized complex lipids in blood plasma. Nat Commun. 2022;13(1):6547.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNi Z, et al. Guiding the choice of informatics software and tools for lipidomics research applications. Nat Methods. 2023;20(2):193\u0026ndash;204.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eConroy MJ, et al. LIPID MAPS: update to databases and tools for the lipidomics community. Nucleic Acids Res. 2024;52(D1):D1677\u0026ndash;82.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAimo L, et al. The SwissLipids knowledgebase for lipid biology. Bioinformatics. 2015;31(17):2860\u0026ndash;6.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCastro-Alves V, Orešič M, Hy\u0026ouml;tyl\u0026auml;inen T. Lipidomics in nutrition research. Curr Opin Clin Nutr Metab Care. 2022;25(5):311\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu Z, et al. Lipidomics: Mass spectrometric and chemometric analyses of lipids. Adv Drug Deliv Rev. 2020;159:294\u0026ndash;307.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBuja LM. Lipid abnormalities in myocardial cell injury. Trends Cardiovasc Med. 1991;1(1):40\u0026ndash;5.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHadas Y, et al. Altering Sphingolipid Metabolism Attenuates Cell Death and Inflammatory Response After Myocardial Infarction. Circulation. 2020;141(11):916\u0026ndash;30.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEmpinado HM, et al. Diaphragm dysfunction in heart failure is accompanied by increases in neutral sphingomyelinase activity and ceramide content. Eur J Heart Fail. 2014;16(5):519\u0026ndash;25.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang XJ, et al. Pharmacological inhibition of arachidonate 12-lipoxygenase ameliorates myocardial ischemia-reperfusion injury in multiple species. Cell Metab. 2021;33(10):2059. \u0026ndash;2075.e10.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMcCluskey ER, et al. The arachidonic acid metabolic capacity of canine myocardium is increased during healing of acute myocardial infarction. Circ Res. 1982;51(6):743\u0026ndash;50.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBilheimer DW, et al. Fatty acid accumulation and abnormal lipid deposition in peripheral and border zones of experimental myocardial infarcts. J Nucl Med. 1978;19(3):276\u0026ndash;83.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHlushchenko I, et al. Linking Variability of Leukocyte Lipid Metabolism to Circulating Lipids, Lipoprotein Composition, and Cardiovascular Risk in the Finnish Adult Population. Arterioscler Thromb Vasc Biol. 2025;45(8):1416\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTokutome M, et al. Peroxisome proliferator-activated receptor-gamma targeting nanomedicine promotes cardiac healing after acute myocardial infarction by skewing monocyte/macrophage polarization in preclinical animal models. Cardiovasc Res. 2019;115(2):419\u0026ndash;31.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRaas Q et al. Metabolic rerouting via SCD1 induction impacts X-linked adrenoleukodystrophy. J Clin Invest, 2021. 131(8).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCavallero S, et al. Exercise mitigates flow recirculation and activates metabolic transducer SCD1 to catalyze vascular protective metabolites. Sci Adv. 2024;10(7):eadj7481.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKrajnik A, et al. Phosphoinositide Signaling and Mechanotransduction in Cardiovascular Biology and Disease. Front Cell Dev Biol. 2020;8:595849.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eManna P, Jain SK. Phosphatidylinositol-3,4,5-triphosphate and cellular signaling: implications for obesity and diabetes. Cell Physiol Biochem. 2015;35(4):1253\u0026ndash;75.\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":"Lipidomics, Myocardial Infarction, Computational Biology, Bioinformatics, Structure-Function Relationship, Lipid Deconstruction, Biomarker Discovery","lastPublishedDoi":"10.21203/rs.3.rs-7238340/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7238340/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eLipiDecipher is a structure-oriented framework designed to overcome the functional interpretation bottleneck in lipidomics. By integrating lipid-protein-pathway mapping with a unique lipid deconstruction algorithm, it connects molecular structure to biological function. Analyzing a myocardial infarction cohort, LipiDecipher revealed novel mechanisms, including a progressive depletion of phosphatidylinositol-associated fatty acids in recurrent patients, which points to deteriorating membrane repair capacity. This powerful paradigm enables the discovery of novel biological insights and robust biomarkers, significantly advancing clinical lipidomics research.\u003c/p\u003e","manuscriptTitle":"LipiDecipher: A Structure-Oriented Framework for Mechanistic Interpretation in Clinical Lipidomics","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-19 11:29:20","doi":"10.21203/rs.3.rs-7238340/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":"f592d20f-4f63-48fe-8a5a-869ef779faee","owner":[],"postedDate":"August 19th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-09-26T10:08:19+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-19 11:29:20","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7238340","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7238340","identity":"rs-7238340","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}