Rapid and robust sex determination from ancient enamel proteomes using protSexInferer

19 20 Accurate biological sex determination of ancient remains is critical for archaeological, 21 anthropological, and forensic studies, but remains challenging for morphologically 22 ambiguous and highly degraded endogenous DNA samples. Paleo-proteomics sex 23 identification approaches, targeting sexually dimorphic amelogenin isoforms (AMELX and 24 AMEL Y), present a promising solution. However, current workflows rely on manual 25 verification of a few specific peptide markers, a process that lacks standardization and is 26 susceptible to false-positive AMEL Y signals. To overcome these limitations, we developed 27 protSexInferer, a lightweight, open-source bioinformatic pipeline for automated sex 28 estimation from paleo-proteomic data. Our method uses the ratio of AMEL Y-specific peptides 29 to all detected AMEL Y- and AMELX-specific peptides (i.e., the RAMEL Y value) rather than the 30 mere presence or absence of AMEL Y signals for sex classification. We demonstrated that the 31 RAMEL Y value clearly distinguishes male and female individuals in both reference and 32 independent validation datasets, enabling reliable sex assignments even in cases where 33 conventional intensity-based comparisons (e.g., AMEL Y-59M vs. AMELX-60) are 34 ambiguous. This ratio-based approach effectively mitigates the impact of false-positive 35 AMEL Y signals, therefore eliminating the need for time-consuming manual verification, and 36 remains reliable even for samples with low peptide yields. Equipped with pre-constructed 37 protein reference databases, protSexInferer provides a robust, standardized, and end-to-end 38 solution for paleo-proteomic sex determination. 39 40

paleo-proteomics, sex determination, automated pipeline, AMELY ratio (RAMEL Y), 41 dental enamel 42 43 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 2

44 45 The accurate determination of the biological sex of ancient samples is a key facet of 46 archaeological and anthropological research, and it is fundamental in the interpretation of 47 sexual dimorphism features(Madupe et al., 2025), ancient demographic 48 variations(García-Fernández et al., 2020; Goldberg et al., 2017; Gretzinger et al., 2022; Haak 49 et al., 2015), and reconstructions of past social structures(Filiatreau, 2019; Fowler et al., 2022; 50 J. Wang et al., 2025; Yüncü et al., 2025). And it is also valuable in forensic research(Dash et 51 al., 2020; Krishan et al., 2016; Mikšík et al., 2023). 52 53 Morphological sex assessment relies on the analysis of osteological sexually dimorphic traits. 54 This method, however, becomes less reliable when the address lacks critical morphological 55 sex-diagnostic elements or features not fully developed(Krishan et al., 2016; MAYS, 2000; 56 WALDRON, 1987). DNA-based molecular techniques determine sex by identifying sex 57 chromosome-specific DNA markers or by quantifying the ratio of sequencing reads mapped 58 to the X and Y chromosomes(Loreille et al., 2018; Mittnik et al., 2016; Skoglund et al., 2013). 59 These approaches depend on the availability of adequately preserved DNA. The accurate 60 sex-determination of remains with absent or undeveloped critical morphological 61 sex-diagnostic features and unavailable DNA preservation is still an issue. 62 63 The development of mass spectrometry techniques, especially LC-MS/MS, to identify 64 proteins in biological samples, enables the wide application of paleo-proteomics in sex 65 identification(Gamble et al., 2024). This approach primarily relies on detecting sex-specific 66 isoforms of the amelogenin gene — specifically, the AMELX, which is expressed on the X 67 chromosome, and its Y -linked counterpart, AMEL Y(Fincham et al., 1991; Parker et al., 2019; 68 Stewart et al., 2017). As the most abundant protein in dental enamel, amelogenin is 69 particularly resistant to degradation and exogenous contamination owing to the highly 70 mineralized nature of enamel(Castiblanco et al., 2015; Demarchi et al., 2016; Mazumder et 71 al., 2014). Since amelogenin is more stable over time compared to proteins or DNA derived 72 from less mineralized biological sources, the analysis of sex-specific amelogenin isoforms is 73 feasible in poorly preserved ancient samples, even in some fragmented million-year-old 74 fossils(Fong-Zazueta et al., 2025; Green et al., 2025; Taurozzi et al., 2024; T. Wang et al., 75 2025). Since it can be applied to subadult, fragmented, and endogenous DNA-free specimens, 76 the paleo-proteomic amelogenin sex estimation has proven to be an effective and robust 77

for sex determination(Buonasera et al., 2020). 78 79 Despite the presence of over 20 amino acid differences between the human AMEL Y isoform 80 and AMELX isoform 3(Parker et al., 2019), current workflows for paleo-proteomic sex 81 estimation still rely on the detection of one (59M) or a few manually selected sexually 82 dimorphic peptides(Stewart et al., 2017). Since the signal of AMELY-specific peptides is 83 regarded as unambiguous evidence for a “female” individual in previous studies, it requires 84 extensive manual verification by researchers specialized in paleo-proteomics to eliminate the 85 risk of false positives AMEL Y signals in mass spectrometry database search(Adair et al., 86 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 3 2025; Cleland et al., 2024; Gowland et al., 2021; Lugli et al., 2019; Wasinger et al., 2019). 87 Consequently, current paleo-proteomic sex estimation suffers from insufficient 88 standardization and a lack of an integrated automatic pipeline, limiting its reproducibility, 89 scalability, and overall robustness. 90 91 To address these limitations, we developed protSexInferer, an open-source Nextflow pipeline 92 for paleo-proteomic sex estimation. Our pipeline determines biological sex by calculating the 93 ratio of the number of AEML Y-specific peptides to all detected AMELX- and 94 AMEL Y-specific amelogenin peptides (RAMEL Y). Sex classification thresholds are established 95 based on the RAMEL Y distribution in known sex samples and validated using independently 96 published paleo-proteomic datasets. Furthermore, our pipeline integrates our pre-constructed 97 amelogenin reference database and can parse the results of various search engines (PEAKS, 98 pFind, MaxQuant, and DIA-NN). Its final results include information on all identified 99 AMELX- and AMEL Y-specific peptides and sex assessment reports. Our results showed that 100 protSexInferer provides a highly accurate, robust, end-to-end, and user-friendly solution for 101 paleo-proteomic sex estimation. 102 103

and discussions 104 105 Workflow 106 107

database construction 108 Before running protSexInferer, appropriate protein reference databases must be constructed or 109 selected. The pipeline provides multiple pre-built databases optimized for different analytical 110 contexts (Supplementary Text). By default, protSexInferer uses the “Hominidae enamel 111 protein database”, including Pongo(Patramanis et al., 2023), Gorilla(Patramanis et al., 2023), 112 Pan(Patramanis et al., 2023), Homo(Patramanis et al., 2023; Welker et al., 2020), 113 Gigantopithecus(Welker et al., 2019), and Paranthropus(Madupe et al., 2025), designed for 114 tooth enamel samples from unidentified primate specimens, with particular relevance to 115 paleoanthropology research. The default enamel protein database comprises the following 116 proteins: AHSG (FETUA), ALB (ALBU), AMBN, AMELX, AMELY , AMTN, COL17A1, 117 ENAM, KLK4, MMP20, ODAM, and TUFT1. In addition, we provide alternative reference 118 databases optimized for Neolithic archaeological samples, modern forensic specimens, and 119 unidentified mammalian enamel samples (Supplementary Text). Users may select and apply 120 the most appropriate reference database according to their specific research context. Using 121 the same standardized reference databases ensures consistency in RAMEL Y value calculations 122 across studies. 123 124 Data input requirements 125 The protSexInferer accepts output files from commonly used protein search engines: PEAKS 126 (v.11 or later; protein-peptides.csv) (Xin et al., 2022), MaxQuant (v.2.6.0.0 or later; 127 evidence.txt) (Cox and Mann, 2008), pFind (v.3.2.1 or later; pFind.proteins) (Chi et al., 2018), 128 and the DDA mode in DIA-NN (v.2.3.0 or later; report.parquet; currently available for Homo 129 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 4 samples only) (Demichev et al., 2020). Database search must be performed by the user prior 130 to running protSexInferer, with parameters optimized for ancient proteomic data as detailed in 131 Supplementary Text. 132 133 Running protSexInferer 134 protSexInferer analysis proceeds through 3 main steps (Figure 1): 135 1. AMELX/AMEL Y-specific peptide classification 136 2. RAMEL Y calculation and confidence interval estimation 137 3. Sex determination and report generation 138 139 Step 1: AMELX/AMELY-specific peptide classification 140 The pipeline first parses database search results from the specified software platform. For 141 non-PEAKS inputs, files are automatically converted to a PEAKS-compatible format. The 142 pipeline removes modification information to obtain the base peptide sequence and merges 143 identical peptide sequences to avoid duplicate counting. Then, the unique peptides matching 144 the amelogenin proteins are extracted. 145 146 Classification of AMELY-specific and AMELX-specific peptides proceeds through two 147 sequential filters. First, based on the database search results, peptides matching exclusively to 148 AMEL Y (with no matches to AMELX) are considered putatively AMEL Y-specific, and vice 149 versa for AMELX-specific peptides. Second, these putatively specific peptides are 150 cross-validated against the complete reference database: any peptide showing identical 151 matches to both AMELX and AMELY sequences—indicating origin from homologous 152 regions—is excluded from quantification. Only peptides that pass both filters are considered 153 authentic AMELX/Y-specific peptides for RAMEL Y calculation. 154 155 Step 2: RAMELY calculation and confidence interval estimation 156 For each sample, the pipeline calculates the RAMELY ratio as: RAMEL Y = nAMEL Y / (nAMEL Y + 157 nAMELX), where nAMELY represents the count of AMEL Y-specific peptides and nAMELX 158 represents the count of AMELX-specific peptides. Under the assumption that nAMEL Y and 159 nAMELX follow a Bernoulli distribution (each detected peptide originates from either the X- or 160 Y-chromosome), the pipeline computes 95% confidence intervals (CI) using the normal 161 approximation: RAMEL Y ± 1.960 × RAMEL Y × (1- RAMEL Y) / (nAMEL Y + nAMELX). 162 163 Step 3: Sex determination and report generation 164 Sex assignment is performed by comparing each sample's RAMEL Y value with established 165 threshold ranges. By default, the pipeline uses pre-calculated thresholds specific to each 166 search software, derived from known-sex reference individuals (Table S3). The female range 167 is defined as RAMEL Y values less than or equal to the upper limit of 95% CI of the observed 168 maximum RAMEL Y in reference females; the male range is defined as RAMEL Y values greater 169 than or equal to the lower limit of 95% CI of the observed minimum RAMEL Y in reference 170 males. Samples with RAMEL Y values falling within one category's range are assigned the 171 corresponding sex; samples with values outside both ranges or overlapping the threshold 172 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 5 boundary are reported as "Unknown Sex (U)". 173 174 To facilitate downstream analysis of amelogenin peptides, the pipeline outputs 175 comprehensive peptide information tables for each sample, listing all filtered amelogenin 176 peptides. For each peptide, the table includes its sequence, classification status 177 (“AMELX-unique”, “AMEL Y-unique”, or “Both”), and the summed “#Spec”, “Intensity”, 178 and “Area” values. These results are accompanied by a RAMELY value distribution plot with 179 female and male threshold ranges shaded and samples colored by their assigned sex, and a 180 final sex assignment report containing sample names, RAMEL Y values with 95% confidence 181 intervals (CI), and corresponding sex determinations. 182 183 184 Figure 1: Overview of the protSexInferer Pipeline 185 186 Customization 187 While protSexInferer provides pre-built reference databases and optimized parameters for 188 ancient proteomic data, users may customize these components when they are unsuitable for 189 specific research contexts. Users can construct custom reference databases, provided that 190 FASTA headers follow the required naming convention (Supplementary Text). When working 191 with custom databases or search parameters, the default RAMEL Y thresholds may not be 192 appropriate. In such cases, users can re-estimate thresholds based on known-sex reference 193 samples. The newly estimated thresholds can then be applied to unknown samples in 194 subsequent analyses, offering flexibility for diverse sex determination experiments. 195 196 The effectiveness of protSexInferer’s sex identification 197 To establish the RAMELY intervals for sex determination, we calculated the RAMEL Y values for 198 76 reference samples of known sex(Lugli et al., 2019; Madupe et al., 2025; Parker et al., 2019; 199 Rebay-Salisbury et al., 2020; Stewart et al., 2017). Detailed information on these reference 200 individuals is listed in Table S1. We observed that, while the absolute RAMEL Y ranges of the 201 same LC-MS/MS raw data varied considerably after processing with different software, all 202 RAMEL Y values consistently formed two distinct clusters corresponding to sample sex (Figure 203 2). The RAMEL Y thresholds for each search engine are provided in Table S3. Comparing across 204 the 4 supported software platforms, we found MaxQuant yielded the lowest thresholds, 205 suggesting more conservative detection of AMEL Y-specific signals with fewer false positives, 206 while DIA-NN (the DDA mode) produced the highest thresholds, indicating more permissive 207 peptide identification. Among all software, PEAKS exhibited the largest separation between 208 the minimum RAMEL Y of males and the maximum RAMELY of females, offering the greatest 209 discriminatory power. Based on these superior performances, we selected PEAKS as the 210 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 6 default search engine and established its corresponding thresholds (male: RAMEL Y > 0.088; 211 female: RAMEL Y < 0.055) for sex determination. 212 213 214 Figure 2: The estimated RAMELY values for known sex reference samples. Va lu es we re 215 calculated from the database search outputs of PEAKS (A), pFind (B), DIA-NN (C), and 216 MaxQuant (D). We have indicated the prior determined sex of the published samples in 217 parentheses after their sample names. The detailed information of these samples can be found 218 in Table S1. We estimate the RAMELY threshold to distinguish males and females based on the 219 RAMEL Y interval for each sex. The blue dashed line indicates the upper limit of 95% CI of the 220 maximum RAMEL Y for female individuals, and the red dashed line indicates the lower limit of 221 95% CI of the minimum RAMELY for male individuals. 222 223 Subsequently, we applied the new pipeline to an independent validation dataset comprising 224 69 individuals whose sex had previously been determined via a proteomic-based method 225 (Demeter et al., 2022; Gowland et al., 2021; Lugli et al., 2019; Madupe et al., 2025; Parker et 226 al., 2019; Tsutaya et al., 2025; Welker et al., 2020), including both adult and non-adult 227 individuals (also including the babies in gestation weeks), deciduous and permanent teeth, 228 archaeological (up to ~2Ma) and present-day specimens, sufficient and in sufficient enamel 229 samples, and healthy and diseased teeth. Detailed information on these validation individuals 230 is listed in Table S2. Our newly assigned sexes were consistent with prior determinations in 231 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 7 almost all cases, regardless of the search engine used (Figure 3). This indicated that the 232 RAMEL Y-based sex identification method is robust to the changes in database search software 233 and specific parameters. Collectively, these results demonstrate that our novel pipeline serves 234 as a robust and generalizable framework for paleo-proteomic sex determination. Notably, the 235 individual JY63, previously morphologically assessed as male, was reclassified as female by 236 our new method. This reclassification aligns with the results of the previous proteomic-based 237 sex determination(Parker et al., 2019). 238 239 240 Figure 3: The estimated RAMELY values for unknown sex validation samples and new 241 reported samples from Xiawanggang Site. Values were calculated from the database search 242 outputs of PEAKS (A), pFind (B), DIA-NN (C), and MaxQuant (D). The samples with 243 RAMEL Y values less than the female RAMEL Y threshold are assigned as females (indicated by 244 blue points), while samples with RAMEL Y values larger than the male RAMELY threshold are 245 identified as males (indicated by red points). 246 247 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 8 248 Figure 4: The estimated RAMELY values for 10 randomly selected individuals calculated 249 from PEAKS database search results using different reference databases. A) Searches 250 using the “Human enamel proteins database” designed for human specimens. B) Searches 251 using the “Hominidae AMEL database” designed for samples with high AMEL peptide 252 content. C) Searches using the “Mammalian enamel proteins database” designed for 253 unidentified mammalian samples. Dashed lines in the figure represent the RAMEL Y thresholds 254 calculated using the default “Hominidae enamel protein database”. Detailed information on 255 each database can be found in Supplementary Text. 256 257 The robustness of protSexInferer to the false positive AEMLY signals and 258 low-data-amount samples 259 The signal of AMEL Y-specific peptides is commonly regarded as evidence of the presence of 260 the Y chromosome, that is, a male individual. However, our results showed that although 261 female individuals should theoretically not carry AMEL Y-specific peptides, AMEL Y-specific 262 peptides were still detected in the database search results of some female individuals and 263 persisted even after we repeatedly adjusted the reference datasets (Figure 4). And the false 264 positive AMEL Y-specific peptide signals have been reported in several cases as well(Parker 265 et al., 2019). These false positive signals also pose a risk of misclassifying females as males 266 in sex determination approach that rely solely on the relative strength of the AMEL Y-59M 267 and AMELX-60 peptides (Madupe et al., 2025). 268 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 9 269 The false positive AMEL Y-specific peptide signals were in our expectation. In standard false 270 discovery rate (FDR) reliability assessments, the “entrapment sequences” are introduced by 271 adding random swapped/transversed mutations into the target human sequences, or by adding 272 homologous proteins from genetically distant species (e.g., mouse or yeast). These 273 “entrapment sequences” are expected to be detected at a ratio lower than the PSM FDR level. 274 However, for the female specimens, as all AMEL Y sequences in the searching database have 275 higher similarity than the standard “entrapment sequences” to the target AMELX sequences, 276 the rate of false-positive AMEL Y identifications is expected to be modestly higher than the 277 nominal PSM FDR in the sex determination analysis (0.010). 278 279 To evaluate the impact of these false positives, we applied the Madupe et al. method to results 280 from 3 of the 4 supported search engines, as pFind was not supported for intensity 281 information. For MaxQuant, which exhibited the lowest false positive AMEL Y rate, only one 282 female individual was incorrectly classified as male. It was consistent with the lowest RAMEL Y 283 thresholds yielded by MaxQuant. However, for PEAKS and DIA-NN (DDA mode), the 284 Madupe et al. method produced numerous false positive male assignments due to their more 285 permissive peptide identification (Figure 5). In contrast, within our newly proposed 286 RAMEL Y-based framework, these misclassified individuals can be reliably reassigned as female 287 (Figure 3), eliminating the need for the time-consuming manual verification of all AMEL Y 288 peptide identifications. 289 290 291 Figure 5: The relationship between the intensity of AMELY-59M and AMELX-60 292 peptides for reference and validation samples. M(R), M(V), F(R), and F(V) represent the 293

of reference male samples, validation male samples, reference female samples, and 294 validation female samples, respectively. The blue vertical dashed line indicates the lowest 295 log2(AMELX-60) value among the reference male individuals. The red horizontal dashed 296 line indicates the lowest log2(AMEL Y-59) value among the reference male individuals. For 297 the MaxQuant results, since the AMEL Y-59 signal was not detected in the JY33 individual, 298 the second-lowest value (from the 16_MdT-1 individual) was used as the reference. The 299 detailed information of all amelogenin peptides for these individuals was provided in the 300 updated ProteomeXchange dataset (identifier: PXD072753). 301 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 10 302 Incorporating the full-length AMEL Y sequence substantially improved the sensitivity and 303 accuracy of sex determination. In some instances where the intensities of 59M-related 304 peptides were markedly low (less than 10% of typical levels) , such as in samples VG3_2 and 305 VG19_2(Gowland et al., 2021), or where the most typical AMEL Y-specific peptides 306 (M(ox)IRPPY , m/z = 396.710; SMIRPPY , m/z = 432.230; SM(ox)IRPPY , m/z = 440.220) was 307 partially or completely absent, additional AMEL Y-specific peptides enabled correct sex 308 assignment. Those peptides would be helpful on badly preserved ancient specimens with no 309 (or very few) 59M-related peptides detected (Figure 5), or on some basal mammals with 310 different amino acids around AMEL Y-59. And the sex determination based on RAMEL Y values 311 could be effective and exact for both adult and non-adult individuals, and for both deciduous 312 and permanent teeth. 313 314 To investigate how the total number of amelogenin peptides influences the estimated RAMELY 315 value in a sample, we evaluated their correlation. The results showed no strong correlation 316 between the RAMEL Y value and the total number of amelogenin peptides (R2 = 0.053, p = 317 0.004, for male individuals; R2 = 0.110, p = 0.007, for female individuals with R AMEL Y value > 318 0; Figure S1). This indicated that, although the absolute numbers of amelogenin peptides 319 fluctuated across ancient samples of varying ages and preservation conditions, the ratio of 320 AEML Y-specific peptides was generally stable. As long as the signal from AMEL Y-specific 321 peptides remains detectable, RAMEL Y values can serve as a reliable reference for sex 322 identification, even in enamel samples with insufficient mineralization or severe degradation. 323 Furthermore, the estimated confidence intervals of RAMEL Y values quantitatively reflected the 324 uncertainty associated with the amount of data. 325 326 Application for sex determination of scattered teeth from Xiawanggang site (XWG) 327 The protSexInferer workflow was then applied for 8 randomly scattered teeth from the 328 Xiawanggang site (Table S2). For all these specimens, the RAMEL Y values calculated from the 329

of 4 search engines consistently exceeded the minimum male threshold (Figure 3), 330 allowing them to be confidently classified as male. 331 332 Sex determination with other databases 333 To examine the effectiveness of the protSexInferer workflow using other databases, 3 334 different databases were tested on 10 randomly selected individuals. All 3 different databases 335 showed large separation between the minimum RAMEL Y of males and the maximum RAMEL Y of 336 females (Figure 4), and the results with databases smaller than the default “Homonidae 337 enamel proteins database”, such as the “Human enamel proteins database” (Figure 4A) and 338 the “Hominidae AMEL database” (Figure 4B), were similar to the default result, and could 339 share the same thresholds. However, results with a larger “Mammalian enamel proteins 340 database” did not fit the default thresholds, as we observed a notably higher false-positive 341 rate (Figure 4C), possibly due to extra match errors introduced by the extra protein sequences, 342 which are phylogenetically distant from the expected sample taxon. Thus, we strongly 343 recommend using custom thresholds with any database larger than the default. 344 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 11 345

of our workflow 346 Our pipeline would be limited by the absence of the AMEL Y gene or expression in 347 males(Mitchell et al., 2006; Turrina et al., 2011), though very rare. It would be limited by the 348 random absence of AMEL Y-specific peptides in the enamel specimens, especially for the 349 highly mineralized samples or insufficient sampling experiments. And this workflow is only 350 applicable to mammalian species that possess both AMELX and AMEL Y isoforms, including 351 primates, but excluding species such as mice, which have only a single isoform of AMEL(X) 352 (Lau et al., 1989; Fincham et al., 1991). 353 354

and materials 355 356 Pipeline framework and portability 357 protSexInferer is developed in the Nextflow programming language(Di Tommaso et al., 358 2017). Its built-in task scheduling and parallelization capabilities make the pipeline 359 particularly suitable for large-scale sample batches. All software dependencies are managed 360 through Nextflow's Conda integration, which automatically handles environment 361 configuration without requiring administrative privileges, ensuring adaptability to diverse 362 computing platforms with minimal setup. 363 364 Sample collection 365 We compiled a diverse enamel sample dataset for pipeline evaluation, comprising 76 samples 366 with known sex and 69 samples with unknown sex from previously published studies (Table 367 S1, Table S2). The 69 unknown-sex samples had prior sex determinations based on 368 alternative proteomic methods. Additionally, we included 8 newly analyzed enamel samples 369 from randomly scattered human teeth excavated from the Xiawanggang site, Henan, China 370 (ca. 5,000 BP) (Henan Institute of Archaeology, 1989), for which sex information was 371 previously undetermined. 372 373 Protein extraction from the new Xiawanggang samples 374 An acid etching method was used to extract protein from tooth enamel(Stewart et al., 2017). 375 Disposable toothbrushes were used to remove surface contaminants from a small area of 376 enamel for etching. At the same time, the remaining teeth were wrapped with parafilm to 377 prevent contact with any liquids. Before etching, the small enamel area was initially washed 378 with 3% H2O2 for 30 seconds, followed by a rinse with ultrapure water. Approximately 100 379 µL of 5% (v/v) HCl was placed in the cap of a 1.5-mL microcentrifuge tube. A 2-minute etch 380 was performed by immersing the etching region in the HCl solution, and the initial etch 381 solution was discarded. A second etch, lasting 15 minutes, was carried out in the cap of 382 another separate microcentrifuge tube, and the etch solution was retained. This second etch 383 was repeated, and the etch solutions were combined and then desalted using C18 ZipTips 384 (Thermo Fisher Scientific). After etching, the etched area was treated with 100 µL of 50 mM 385 ammonium bicarbonate solution for 1 minute to neutralize the acid. It was then rinsed with 386 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 12 ultrapure water for 30 seconds and dried. The protein peptides were eluted into a solution of 387 0.1% trifluoroacetic acid (TFA) and 80% acetonitrile (ACN). The peptide mixture was further 388 divided into different aliquots and dried for Liquid chromatography-tandem mass 389 spectrometry (LC-MS/MS) analysis. All sample preparation for the experiment was 390 conducted in the dedicated clean room at the Molecular Paleontology Laboratory, IVPP of the 391 Chinese Academy of Sciences in Beijing. 392 393 Mass spectrometry data acquisition of the new Xiawanggang samples 394 The LC-MS/MS experiment was done with an Orbitrap Exploris 480 (Thermo) interfaced 395 with an Easy n-LC 1200 HPLC system (Thermo) at Fudan University, Shanghai, China. The 396 peptides were separated on a 75 μ m id×25 cm analytical column, which was packed in-house 397 using reversed-phase silica of 1.9 μ m (Reprosil-Pur C18 AQ, Dr. Maisch GmbH). Buffer A 398 was 0.1% formic acid in water, and Buffer B was 80% acetonitrile and 0.1% formic acid. An 399 80 min gradient was used with the following profile: 5-8% B, 2 min, at a flow rate of 200 400 nL/min; 8-44% B, 38 min, 200 nL/min; 44-70% B, 8 min, 200 nL/min; 70-100% B, 2min, 401 200 nL/min; 100% B, 10 min, 200 nL/min; 100-5% B, 2 min, 200 nL/min; 5% B, 2 min, 300 402 nL/min; 5-100% B, 6 min, 300 nl/min; 100% B, 10 min, 300 nL/min. Full MS scans were 403 acquired for the first 65 min, after which the column was washed and re-equilibrated for 15 404 min without data acquisition. The full MS data acquisition was conducted across the range of 405 m/z 350–1600, with a resolution of 60k at m/z 200. The AGC target was set to “Standard,” 406 and the maximum injection time mode was set to “Auto”. The MS/MS spectra were acquired 407 with a resolution of 15k at m/z 200, a maximum injection time of 30 ms, and a normalized 408 collision energy of 30%. The AGC target was also set to “Standard”. 409 410 Database search strategy 411 It is described in detail in Supplementary Text. The parameters were set to the default. 412 413

414 In summary, our protSexInferer pipeline provides an automated approach for paleo-proteomic 415 sex determination based on the RAMELY ratio—the proportion of AMEL Y-specific peptides 416 relative to total AMELX- and AMEL Y-specific peptides. By leveraging peptide ratios rather 417 than binary signal detection, the method effectively eliminates false-positive AMEL Y 418 identifications and obviates time-consuming manual verification. Unlike approaches targeting 419 limited diagnostic sites (e.g., 59M), the pipeline incorporates all variant sites across AMELX 420 and AMEL Y isoforms, fully utilizing available peptide data and maintaining high sensitivity 421 even for low-quality samples. We showed that the RAMEL Y values consistently discriminate 422 sex across published reference and validation datasets. Compared to intensity-based methods, 423 the ratio-based framework exhibits greater robustness against sporadic erroneous 424 identifications. With integrated standardized databases and support for multiple search engine 425 outputs, protSexInferer offers a robust, user-friendly, end-to-end solution for paleo-proteomic 426 sex estimation. 427 428 Conflict of interest: The authors declare that they have no conflict of interest. 429 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 13 430 Acknowledgments 431 We thank our anonymous reviewers for their helpful comments. This study was supported by 432 the Research Program of Chinese Academy of Sciences (XDA0460305), the National Science 433 Foundation of China (32293192), the Chinese Academy of Sciences (CAS) (YSBR-019), and 434 the Archaeological Talent Promotion Program of China (2024-278). 435 436 Author contributions 437 Q.F. designed and supervised the research project. Z.W. processed the raw data. F.B. and Z.W. 438 wrote the code, analyzed, and visualized the data. S.X. prepared the archaeological samples 439 and materials. F.B., Z.W., and Q.F. did the data investigation, wrote the manuscript, discussed 440 and revised the manuscript. Z.W . and F.B. wrote and prepared the supplementary materials. 441 All authors discussed, critically revised, and approved the final version of the manuscript. 442 443 Code and data availability 444 The source code of the protSexInferer pipeline is available on GitHub at 445 https://github.com/QFuLab/protSexInferer with a manual. LC-MS/MS raw data, the 446 estimated RAMELY values, and filtered amelogenin peptides have been uploaded to the 447 ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) with the data 448 set identifier: PXD072753. 449 450 Reviewer access details: Log in to the PRIDE website using the following details: Project 451 accession: PXD072753; Token: svedVIScJcs9; Alternatively, the reviewer can access the 452 dataset by logging in to the PRIDE website using the following account details: Username: 453 [email protected]; Password: oTDrhEs44J3Z. 454 455

456 Adair, L.R., Lewis, M.E., Collins, M.J., Cramer, R., 2025. LAP-MALDI MS analysis of 457 amelogenin from teeth for biological sex estimation. J Pharm Biomed Anal 255, 116599. 458 https://doi.org/10.1016/j.jpba.2024.116599 459 Buonasera, T., Eerkens, J., de Flamingh, A., Engbring, L., Yip, J., Li, H., Haas, R., 460 DiGiuseppe, D., Grant, D., Salemi, M., Nijmeh, C., Arellano, M., Leventhal, A., Phinney, 461 B., Byrd, B.F., Malhi, R.S., Parker, G., 2020. A comparison of proteomic, genomic, and 462 osteological methods of archaeological sex estimation. Sci Rep 10, 11897. 463 https://doi.org/10.1038/s41598-020-68550-w 464 Castiblanco, G.A., Rutishauser, D., Ilag, L.L., Martignon, S., Castellanos, J.E., Mejía, W ., 465 2015. Identification of proteins from human permanent erupted enamel. European 466 Journal of Oral Sciences 123, 390–395. https://doi.org/10.1111/eos.12214 467 Chi, H., Liu, C., Yang, H., Zeng, W.-F., Wu, L., Zhou, W.-J., Wang, R.-M., Niu, X.-N., Ding, 468 Y .-H., Zhang, Y ., Wang, Z.-W., Chen, Z.-L., Sun, R.-X., Liu, T., Tan, G.-M., Dong, 469 M.-Q., Xu, P., Zhang, P.-H., He, S.-M., 2018. Comprehensive identification of peptides 470 in tandem mass spectra using an efficient open search engine. Nat Biotechnol 36, 471 1059–1061. https://doi.org/10.1038/nbt.4236 472 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 14 Cleland, T.P., McGuire, S.A., Beatrice, J.S., Moran, K.S., France, C.A.M., 2024. SPEED-E: A 473 modified version of the sample preparation by Easy extraction and Digestion(-free) 474 protocol for enamel-based sex estimation in archaeological remains. Journal of 475 Archaeological Science 168, 106006. https://doi.org/10.1016/j.jas.2024.106006 476 Cox, J., Mann, M., 2008. MaxQuant enables high peptide identification rates, individualized 477 p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol 478 26, 1367–1372. https://doi.org/10.1038/nbt.1511 479 Dash, H.R., Rawat, N., Das, S., 2020. Alternatives to amelogenin markers for sex 480 determination in humans and their forensic relevance. Mol Biol Rep 47, 2347–2360. 481 https://doi.org/10.1007/s11033-020-05268-y 482 Demarchi, B., Hall, S., Roncal-Herrero, T., Freeman, C.L., Woolley, J., Crisp, M.K., Wilson, 483 J., Fotakis, A., Fischer, R., Kessler, B.M., Rakownikow Jersie-Christensen, R., Olsen, 484 J.V ., Haile, J., Thomas, J., Marean, C.W., Parkington, J., Presslee, S., Lee-Thorp, J., 485 Ditchfield, P., Hamilton, J.F., Ward, M.W., Wang, C.M., Shaw, M.D., Harrison, T., 486 Domínguez-Rodrigo, M., MacPhee, R.D., Kwekason, A., Ecker, M., Kolska Horwitz, L., 487 Chazan, M., Kröger, R., Thomas-Oates, J., Harding, J.H., Cappellini, E., Penkman, K., 488 Collins, M.J., 2016. Protein sequences bound to mineral surfaces persist into deep time. 489 eLife 5, e17092. https://doi.org/10.7554/eLife.17092 490 Demeter, F., Zanolli, C., Westaway, K.E., Joannes-Boyau, R., Duringer, P., Morley, M.W., 491 Welker, F., Rüther, P.L., Skinner, M.M., McColl, H., Gaunitz, C., Vinner, L., Dunn, T.E., 492 Olsen, J.V ., Sikora, M., Ponche, J.-L., Suzzoni, E., Frangeul, S., Boesch, Q., Antoine, 493 P.-O., Pan, L., Xing, S., Zhao, J.-X., Bailey, R.M., Boualaphane, S., Sichanthongtip, P ., 494 Sihanam, D., Patole-Edoumba, E., Aubaile, F., Crozier, F., Bourgon, N., Zachwieja, A., 495 Luangkhoth, T., Souksavatdy, V ., Sayavongkhamdy, T., Cappellini, E., Bacon, A.-M., 496 Hublin, J.-J., Willerslev, E., Shackelford, L., 2022. A Middle Pleistocene Denisovan 497 molar from the Annamite Chain of northern Laos. Nat Commun 13, 2557. 498 https://doi.org/10.1038/s41467-022-29923-z 499 Demichev, V ., Messner, C.B., V ernardis, S.I., Lilley, K.S., Ralser, M., 2020. DIA-NN: neural 500 networks and interference correction enable deep proteome coverage in high throughput. 501 Nat Methods 17, 41–44. https://doi.org/10.1038/s41592-019-0638-x 502 Di Tommaso, P., Chatzou, M., Floden, E.W., Barja, P.P., Palumbo, E., Notredame, C., 2017. 503 Nextflow enables reproducible computational workflows. Nat Biotechnol 35, 316–319. 504 https://doi.org/10.1038/nbt.3820 505 Filiatreau, C., 2019. Kinship-based social inequality in Bronze Age Europe. Science 12. 506 Fincham, A.G., Bessem, C.C., Lau, E.C., Pavlova, Z., Shuler, C., Slavkin, H.C., Snead, M.L., 507 1991. Human developing enamel proteins exhibit a sex-linked dimorphism. Calcif 508 Tissue Int 48, 288–290. https://doi.org/10.1007/BF02556382 509 Fong-Zazueta, R., Krueger, J., Alba, D.M., Aymerich, X., Beck, R.M.D., Cappellini, E., 510 Carrillo-Martin, G., Cirilli, O., Clark, N., Cornejo, O.E., Farh, K.K.-H., Ferrández-Peral, 511 L., Juan, D., Kelley, J.L., Kuderna, L.F.K., Little, J., Orkin, J.D., Paterson, R.S., Pawar, 512 H., Marques-Bonet, T., Lizano, E., 2025. Phylogenetic Signal in Primate Tooth Enamel 513 Proteins and its Relevance for Paleoproteomics. Genome Biol Evol 17, evaf007. 514 https://doi.org/10.1093/gbe/evaf007 515 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 15 Fowler, C., Olalde, I., Cummings, V ., Armit, I., Büster, L., Cuthbert, S., Rohland, N., 516 Cheronet, O., Pinhasi, R., Reich, D., 2022. A high-resolution picture of kinship practices 517 in an Early Neolithic tomb. Nature 601, 584–587. 518 https://doi.org/10.1038/s41586-021-04241-4 519 Gamble, J.A., Spicer, V ., Hunter, M., Lao, Y ., Hoppa, R.D., Pedersen, D.D., Wilkins, J.A., 520 Zahedi, R.P., 2024. Advancing sex estimation from amelogenin: Applications to 521 archaeological, deciduous, and fragmentary dental enamel. Journal of Archaeological 522 Science: Reports 54, 104430. https://doi.org/10.1016/j.jasrep.2024.104430 523 García-Fernández, C., Font-Porterias, N., Kuč inskas, V ., Sukarova-Stefanovska, E., Pamjav, 524 H., Makukh, H., Dobon, B., Bertranpetit, J., Netea, M.G., Calafell, F., Comas, D., 2020. 525 Sex-biased patterns shaped the genetic history of Roma. Sci Rep 10, 14464. 526 https://doi.org/10.1038/s41598-020-71066-y 527 Goldberg, A., Günther, T., Rosenberg, N.A., Jakobsson, M., 2017. Ancient X chromosomes 528 reveal contrasting sex bias in Neolithic and Bronze Age Eurasian migrations. 529 Proceedings of the National Academy of Sciences 114, 2657–2662. 530 https://doi.org/10.1073/pnas.1616392114 531 Gowland, R., Stewart, N.A., Crowder, K.D., Hodson, C., Shaw, H., Gron, K.J., Montgomery, 532 J., 2021. Sex estimation of teeth at different developmental stages using dimorphic 533 enamel peptide analysis. Am J Phys Anthropol 174, 859–869. 534 https://doi.org/10.1002/ajpa.24231 535 Green, D.R., Uno, K.T., Miller, E.R., Feibel, C.S., Aoron, E.E., Beck, C.C., Grossman, A., 536 Kirera, F.M., Kirinya, M.M., Leakey, L.N., Liutkus-Pierce, C., Manthi, F.K., Ndiema, 537 E.K., Nengo, I.O., Nyete, C., Rowan, J., Russo, G.A., Sanders, W.J., Smiley, T.M., 538 Princehouse, P., Vitek, N.S., Cleland, T.P., 2025. Eighteen million years of diverse 539 enamel proteomes from the East African Rift. Nature 643, 712–718. 540 https://doi.org/10.1038/s41586-025-09040-9 541 Gretzinger, J., Sayer, D., Justeau, P., Altena, E., Pala, M., Dulias, K., Edwards, C.J., Jodoin, 542 S., Lacher, L., Sabin, S., Vågene, Å.J., Haak, W., Ebenesersdóttir, S.S., Moore, K.H.S., 543 Radzeviciute, R., Schmidt, K., Brace, S., Bager, M.A., Patterson, N., Papac, L., 544 Broomandkhoshbacht, N., Callan, K., Harney, É., Iliev, L., Lawson, A.M., Michel, M., 545 Stewardson, K., Zalzala, F., Rohland, N., Kappelhoff-Beckmann, S., Both, F., Winger, 546 D., Neumann, D., Saalow, L., Krabath, S., Beckett, S., V an Twest, M., Faulkner, N., 547 Read, C., Barton, T., Caruth, J., Hines, J., Krause-Kyora, B., Warnke, U., Schuenemann, 548 V .J., Barnes, I., Dahlström, H., Clausen, J.J., Richardson, A., Popescu, E., Dodwell, N., 549 Ladd, S., Phillips, T., Mortimer, R., Sayer, F., Swales, D., Stewart, A., Powlesland, D., 550 Kenyon, R., Ladle, L., Peek, C., Grefen-Peters, S., Ponce, P., Daniels, R., Spall, C., 551 Woolcock, J., Jones, A.M., Roberts, A.V ., Symmons, R., Rawden, A.C., Cooper, A., Bos, 552 K.I., Booth, T., Schroeder, H., Thomas, M.G., Helgason, A., Richards, M.B., Reich, D., 553 Krause, J., Schiffels, S., 2022. The Anglo-Saxon migration and the formation of the 554 early English gene pool. Nature 610, 112–119. 555 https://doi.org/10.1038/s41586-022-05247-2 556 Haak, W ., Lazaridis, I., Patterson, N., Rohland, N., Mallick, S., Llamas, B., Brandt, G., 557 Nordenfelt, S., Harney, E., Stewardson, K., Fu, Q., Mittnik, A., Bánffy, E., Economou, 558 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 16 C., Francken, M., Friederich, S., Pena, R.G., Hallgren, F., Khartanovich, V ., Khokhlov, 559 A., Kunst, M., Kuznetsov, P ., Meller, H., Mochalov, O., Moiseyev, V ., Nicklisch, N., 560 Pichler, S.L., Risch, R., Rojo Guerra, M.A., Roth, C., Szécsényi-Nagy, A., Wahl, J., 561 Meyer, M., Krause, J., Brown, D., Anthony, D., Cooper, A., Alt, K.W., Reich, D., 2015. 562 Massive migration from the steppe was a source for Indo-European languages in Europe. 563 Nature 522, 207–211. https://doi.org/10.1038/nature14317 564 Henan Institute of Archaeology, 1989. Xiawanggang in Xichuan (in Chinese). Cultural Relics 565 Publishing House Press, Beijing. 566 Krishan, K., Chatterjee, P.M., Kanchan, T., Kaur, S., Baryah, N., Singh, R.K., 2016. A review 567 of sex estimation techniques during examination of skeletal remains in forensic 568 anthropology casework. Forensic Science International 261, 165.e1-165.e8. 569 https://doi.org/10.1016/j.forsciint.2016.02.007 570 Lau, E.C., Mohandas, T.K., Shapiro, L.J., Slavkin, H.C., Snead, M.L., 1989. Human and 571 mouse amelogenin gene loci are on the sex chromosomes. Genomics 4, 162–168. 572 https://doi.org/10.1016/0888-7543(89)90295-4 573 Loreille, O., Ratnayake, S., Bazinet, A.L., Stockwell, T.B., Sommer, D.D., Rohland, N., 574 Mallick, S., Johnson, P.L.F., Skoglund, P., Onorato, A.J., Bergman, N.H., Reich, D., 575 Irwin, J.A., 2018. Biological Sexing of a 4000-Year-Old Egyptian Mummy Head to 576 Assess the Potential of Nuclear DNA Recovery from the Most Damaged and Limited 577 Forensic Specimens. Genes 9, 135. https://doi.org/10.3390/genes9030135 578 Lugli, F., Di Rocco, G., Vazzana, A., Genovese, F., Pinetti, D., Cilli, E., Carile, M.C., 579 Silvestrini, S., Gabanini, G., Arrighi, S., Buti, L., Bortolini, E., Cipriani, A., Figus, C., 580 Marciani, G., Oxilia, G., Romandini, M., Sorrentino, R., Sola, M., Benazzi, S., 2019. 581 Enamel peptides reveal the sex of the Late Antique ‘Lovers of Modena.’ Sci Rep 9, 582 13130. https://doi.org/10.1038/s41598-019-49562-7 583 Madupe, P.P ., Koenig, C., Patramanis, I., Rüther, P .L., Hlazo, N., Mackie, M., Tawane, M., 584 Krueger, J., Taurozzi, A.J., Troché, G., Kibii, J., Pickering, R., Dickinson, M.R., Sahle, 585 Y ., Kgotleng, D., Musiba, C., Manthi, F., Bell, L., DuPlessis, M., Gilbert, C., Zipfel, B., 586 Kuderna, L.F.K., Lizano, E., Welker, F., Kyriakidou, P., Cox, J., Mollereau, C., Tokarski, 587 C., Blackburn, J., Ramos-Madrigal, J., Marques-Bonet, T., Penkman, K., Zanolli, C., 588 Schroeder, L., Racimo, F., Olsen, J.V ., Ackermann, R.R., Cappellini, E., 2025. Enamel 589 proteins reveal biological sex and genetic variability in southern African Paranthropus. 590 Science 388, 969–973. https://doi.org/10.1126/science.adt9539 591 MAYS, S., 2000. Sex determination in skeletal remains. Human Osteology in Archeology and 592 Forensic Science 117–130. 593 Mazumder, P., Prajapati, S., Lokappa, S.B., Gallon, V ., Moradian-Oldak, J., 2014. Analysis of 594 co-assembly and co-localization of ameloblastin and amelogenin. Front. Physiol. 5. 595 https://doi.org/10.3389/fphys.2014.00274 596 Mikšík, I., Morvan, M., Brů žek, J., 2023. Peptide analysis of tooth enamel – A sex estimation 597 tool for archaeological, anthropological, or forensic research. Journal of Separation 598 Science 46, 2300183. https://doi.org/10.1002/jssc.202300183 599 Mitchell, R.J., Kreskas, M., Baxter, E., Buffalino, L., Van Oorschot, R.A.H., 2006. An 600 investigation of sequence deletions of amelogenin (AMEL Y), a Y-chromosome locus 601 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 17 commonly used for gender determination. Annals of Human Biology 33, 227–240. 602 https://doi.org/10.1080/03014460600594620 603 Mittnik, A., Wang, C.-C., Svoboda, J., Krause, J., 2016. A Molecular Approach to the Sexing 604 of the Triple Burial at the Upper Paleolithic Site of Dolní Vě stonice. PLOS ONE 11, 605 e0163019. https://doi.org/10.1371/journal.pone.0163019 606 Parker, G.J., Yip, J.M., Eerkens, J.W., Salemi, M., Durbin-Johnson, B., Kiesow, C., Haas, R., 607 Buikstra, J.E., Klaus, H., Regan, L.A., Rocke, D.M., Phinney, B.S., 2019. Sex estimation 608 using sexually dimorphic amelogenin protein fragments in human enamel. Journal of 609 Archaeological Science 101, 169–180. https://doi.org/10.1016/j.jas.2018.08.011 610 Patramanis, I., Ramos-Madrigal, J., Cappellini, E., Racimo, F., 2023. PaleoProPhyler: a 611 reproducible pipeline for phylogenetic inference using ancient proteins. Peer 612 Community Journal 3. https://doi.org/10.24072/pcjournal.344 613 Rebay-Salisbury, K., Janker, L., Pany-Kucera, D., Schuster, D., Spannagl-Steiner, M., 614 Waltenberger, L., Salisbury, R.B., Kanz, F., 2020. Child murder in the Early Bronze Age: 615 proteomic sex identification of a cold case from Schleinbach, Austria. Archaeol 616 Anthropol Sci 12, 265. https://doi.org/10.1007/s12520-020-01199-8 617 Skoglund, P ., Storå, J., Götherström, A., Jakobsson, M., 2013. Accurate sex identification of 618 ancient human remains using DNA shotgun sequencing. Journal of Archaeological 619 Science 40, 4477–4482. https://doi.org/10.1016/j.jas.2013.07.004 620 Stewart, N.A., Gerlach, R.F., Gowland, R.L., Gron, K.J., Montgomery, J., 2017. Sex 621 determination of human remains from peptides in tooth enamel. Proc Natl Acad Sci U S 622 A 114, 13649–13654. https://doi.org/10.1073/pnas.1714926115 623 Taurozzi, A.J., Rüther, P.L., Patramanis, I., Koenig, C., Sinclair Paterson, R., Madupe, P.P., 624 Harking, F.S., Welker, F., Mackie, M., Ramos-Madrigal, J., Olsen, J.V ., Cappellini, E., 625 2024. Deep-time phylogenetic inference by paleoproteomic analysis of dental enamel. 626 Nat Protoc 19, 2085–2116. https://doi.org/10.1038/s41596-024-00975-3 627 Tsutaya, T., Sawafuji, R., Taurozzi, A.J., Fagernäs, Z., Patramanis, I., Troché, G., Mackie, M., 628 Gakuhari, T., Oota, H., Tsai, C.-H., Olsen, J.V ., Kaifu, Y ., Chang, C.-H., Cappellini, E., 629 Welker, F., 2025. A male Denisovan mandible from Pleistocene Taiwan. Science 388, 630 176–180. https://doi.org/10.1126/science.ads3888 631 Turrina, S., Filippini, G., V oglino, G., De Leo, D., 2011. Two additional reports of deletion on 632 the short arm of the Y chromosome. Forensic Science International: Genetics 5, 242–246. 633 https://doi.org/10.1016/j.fsigen.2010.10.015 634 WALDRON, T., 1987. The Relative Survival of the Human Skeleton/i5 : Implications for 635 Palaeopathology. Death, Decay and Reconstruction/i5 : Approaches to Archaeology and 636 Forensic Science 55–64. 637 Wang, J., Yan, S., Li, Z., Zan, J., Zhao, Y ., Zhao, J., Chen, K., Wang, X., Ji, T., Zhang, C., 638 Yang, T., Zhang, T., Qiao, R., Guo, M., Rao, Z., Zhang, J., Wang, G., Ran, Z., Duan, C., 639 Zhang, F., Song, Y ., Wu, X., Mace, R., Sun, B., Pang, Y ., Huang, Y ., Zhang, H., Ning, C., 640 2025. Ancient DNA reveals a two-clanned matrilineal community in Neolithic China. 641 Nature 643, 1304–1311. https://doi.org/10.1038/s41586-025-09103-x 642 Wang, T., Y ang, M.A., Zhu, Z., Ma, M., Shi, H., Speidel, L., Min, R., Y uan, H., Jiang, Z., Hu, 643 C., Li, X., Zhao, D., Bai, F., Cao, P., Liu, F., Dai, Q., Feng, X., Yang, R., Wu, X., Liu, X., 644 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint 18 Zhang, M., Ping, W., Liu, Y ., Wan, Y ., Yang, F., Zhou, R., Kang, L., Dong, G., Stoneking, 645 M., Fu, Q., 2025. Prehistoric genomes from Yunnan reveal ancestry related to Tibetans 646 and Austroasiatic speakers. Science 388, eadq9792. 647 https://doi.org/10.1126/science.adq9792 648 Wasinger, V .C., Curnoe, D., Bustamante, S., Mendoza, R., Shoocongdej, R., Adler, L., Baker, 649 A., Chintakanon, K., Boel, C., Tacon, P.S.C., 2019. Analysis of the Preserved Amino 650 Acid Bias in Peptide Profiles of Iron Age Teeth from a Tropical Environment Enable 651 Sexing of Individuals Using Amelogenin MRM. PROTEOMICS 19, 1800341. 652 https://doi.org/10.1002/pmic.201800341 653 Welker, F., Ramos-Madrigal, J., Gutenbrunner, P., Mackie, M., Tiwary, S., Rakownikow 654 Jersie-Christensen, R., Chiva, C., Dickinson, M.R., Kuhlwilm, M., de Manuel, M., 655 Gelabert, P., Martinón-Torres, M., Margvelashvili, A., Arsuaga, J.L., Carbonell, E., 656 Marques-Bonet, T., Penkman, K., Sabidó, E., Cox, J., Olsen, J.V ., Lordkipanidze, D., 657 Racimo, F., Lalueza-Fox, C., Bermúdez de Castro, J.M., Willerslev, E., Cappellini, E., 658 2020. The dental proteome of Homo antecessor. Nature 580, 235–238. 659 https://doi.org/10.1038/s41586-020-2153-8 660 Welker, F., Ramos-Madrigal, J., Kuhlwilm, M., Liao, W., Gutenbrunner, P., de Manuel, M., 661 Samodova, D., Mackie, M., Allentoft, M.E., Bacon, A.-M., Collins, M.J., Cox, J., 662 Lalueza-Fox, C., Olsen, J.V ., Demeter, F., Wang, W., Marques-Bonet, T., Cappellini, E., 663 2019. Enamel proteome shows that Gigantopithecus was an early diverging pongine. 664 Nature 576, 262–265. https://doi.org/10.1038/s41586-019-1728-8 665 Xin, L., Qiao, R., Chen, X., Tran, H., Pan, S., Rabinoviz, S., Bian, H., He, X., Morse, B., 666 Shan, B., Li, M., 2022. A streamlined platform for analyzing tera-scale DDA and DIA 667 mass spectrometry data enables highly sensitive immunopeptidomics. Nat Commun 13, 668 3108. https://doi.org/10.1038/s41467-022-30867-7 669 Yüncü, E., Doğ u, A.K., Kaptan, D., Kılıç, M.S., Mazzucato, C., Güler, M.N., Eker, E., 670 Katırcıoğ lu, B., Chyleń ski, M., Vural, K.B., Sağ lıcan, E., Atağ , G., Bozkurt, D., Pearson, 671 J., Sevkar, A., Altınış ık, N.E., Milella, M., Karamurat, C., Aktürk, Ş ., Y urttaş , E.D., 672 Yıldız, N., Koptekin, D., Yorulmaz, S., Kazancı, D.D., Aydoğ an, A., Gürün, K., 673 Schotsmans, E.M.J., Anvari, J., Rosenstock, E., Byrnes, J., Biehl, P.F., Orton, D., 674 Lagerholm, V .K., Gemici, H.C., Vasic, M., Marciniak, A., Atakuman, Ç., Erdal, Y .S., 675 Kırdök, E., Pilloud, M., Larsen, C.S., Haddow, S.D., Götherström, A., Knüsel, C.J., Özer, 676 F., Hodder, I., Somel, M., 2025. Female lineages and changing kinship patterns in 677 Neolithic Çatalhöyük. Science 388, eadr2915. https://doi.org/10.1126/science.adr2915 678 679 .CC-BY 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted March 25, 2026. ; https://doi.org/10.64898/2026.03.23.713598doi: bioRxiv preprint