Proteomics researchers rely on tandem mass spectrometry to identify the peptides present in a biological sample, but one of the field’s most stubborn challenges has long been the detection of peptides carrying unexpected post-translational modifications. A newly developed computational framework called POMS promises to change that. In a study published in BMC Bioinformatics, Younghee Seo, Eunok Paek and Seungjin Na describe a modification-aware approach to open spectral library searching that substantially increases the number of modified peptides that can be confidently identified from large-scale mass spectrometry datasets, all while preserving the computational efficiency that makes such searches practical in routine laboratory workflows.
Spectral library searching is one of the most sensitive strategies available for peptide identification. Rather than comparing an experimental tandem mass spectrum against theoretically predicted fragments, a library search compares the query spectrum against a curated collection of previously observed, empirically validated spectra. Because real spectra capture the idiosyncrasies of fragmentation behavior, ion intensities and instrument-specific artifacts, library-based matches are typically more reliable and more sensitive than theoretical database searches. This advantage, however, comes with a limitation: the library only contains spectra for the peptide forms that were catalogued when it was built. If a peptide in the sample carries a modification that was not represented in the library, the corresponding spectrum will be absent and the peptide will go unidentified.
Open modification search was developed precisely to close that gap. Instead of restricting the search to a narrow window of precursor mass differences, an open search tolerates arbitrarily large mass shifts between the query precursor and candidate library entries, allowing a modified query spectrum to be matched against the unmodified spectrum of the same peptide sequence, with the mass difference then interpreted as the modification. The strategy has proven effective for discovering unexpected post-translational modifications, but it introduces a subtle and problematic side effect. When a modification is attached to a peptide, it does not only shift the precursor mass; it can also shift the masses of the fragment ions that contain the modified residue. Those fragment-ion shifts distort the spectral similarity calculations at the heart of library matching, degrading alignment between the query and library spectra and causing true matches to score poorly against incorrect candidates.
The authors of the new study identified this failure mode as the central weakness of existing open spectral library search methods. During the candidate retrieval stage, in which the search engine must decide which library spectra deserve detailed scoring, conventional approaches do not adequately account for modification-induced fragment-ion shifts. The consequence is that genuinely modified peptides may never even reach the scoring stage, because their distorted spectra fail to rank the correct library entry among the top candidates. The problem is especially acute for peptides carrying modifications near the C-terminus, where a large fraction of fragment ions in typical fragmentation modes include the modified site and are therefore displaced by the mass shift. For these species, conventional candidate retrieval is particularly susceptible to losing the true match in the noise.
POMS tackles this weakness by changing how candidate spectra are represented and compared. The framework integrates complementary spectral representations with sequence-derived theoretical fragment features, producing a composite description of each library spectrum that is robust to the displacement of fragment ions caused by a modification. Because the theoretical features are derived from the peptide sequence itself, they provide a stable anchor for alignment even when the experimental spectrum’s observed fragments have been shifted by an unanticipated mass. This dual-representation strategy compensates for the modification-induced shifts during alignment, improves the correspondence between query and library spectra, and increases robustness to variability in the position of the modification site along the peptide backbone.
To assess the practical impact of these design choices, the researchers benchmarked POMS on large-scale human tandem mass spectrometry datasets, including data derived from human embryonic kidney cells. The results were striking. During the open search phase, POMS yielded up to approximately eight percent more peptide identifications than conventional candidate retrieval methods, a substantial gain in a field where every additional identification represents hard-won biological information. Cross-validation against independent database search engines, a standard approach for verifying that newly reported matches are credible rather than artifacts of a particular algorithm, demonstrated a greater than five percent increase in consistent peptide identifications, indicating that the additional matches reported by POMS are corroborated by orthogonal search strategies.
The gains were most pronounced for the peptide classes that conventional methods struggle with most. POMS substantially improved the identification of peptides carrying C-terminal modifications, exactly the category for which modification-induced fragment-ion shifts are most disruptive to conventional candidate retrieval. This matters because many biologically important modification events occur at or near peptide termini, and their systematic under-detection can skew downstream biological interpretation. By recovering these difficult cases, POMS expands the visible inventory of the proteome and reduces a form of detection bias that has been embedded in open search workflows.
An important consideration for any computational tool intended for large-scale proteomics is whether its added sensitivity comes at an unacceptable computational cost. The authors emphasize that POMS incorporates modification-aware candidate retrieval while preserving computational efficiency, making it a practical option for the very large datasets that characterize modern proteomics experiments. The framework is designed to slot into existing spectral library search workflows rather than requiring laboratories to rebuild their pipelines from scratch, lowering the barrier to adoption. In addition, the software exploits approximate nearest neighbor techniques to manage the scale of library searching, ensuring that the richer spectral representations do not translate into prohibitive search times.
For the broader proteomics community, the implications extend beyond a simple sensitivity bump. Post-translational modifications govern nearly every aspect of protein function, from enzyme activation and signal transduction to protein degradation and disease states, and mass spectrometry remains the dominant technology for mapping them at scale. Open modification search has been a transformative idea, but its full potential has been throttled by the candidate-retrieval bottleneck that POMS directly addresses. By making the earliest and most consequential step of the search modification-aware, the framework increases both the sensitivity and the reliability of modification-tolerant searching, offering researchers a sharper lens for PTM discovery in contexts ranging from basic cell biology to biomarker research.
The tool is publicly available to the research community under a CC BY-NC-SA 4.0 license through the GitHub repository maintained by the Korea Basic Science Institute, allowing laboratories to evaluate it against their own data and integrate it into their analysis pipelines. The work was supported by the Korea Basic Science Institute and by grants from the National Research Foundation of Korea funded by the Korean Ministry of Science and ICT, reflecting a sustained national investment in proteome informatics. As spectral libraries continue to grow and instruments generate ever-larger volumes of tandem mass spectra, methods such as POMS that combine domain-aware algorithm design with practical efficiency are likely to define the next generation of peptide identification tools, helping researchers extract more complete and more accurate pictures of the modified proteome from the data they already collect.
Subject of Research: Modification-aware open spectral library search for identification of post-translationally modified peptides in tandem mass spectrometry
Article Title: POMS enhances open spectral library search for identification of modified peptides
Article References: Seo, Y., Paek, E., & Na, S. (2026). POMS enhances open spectral library search for identification of modified peptides. BMC Bioinformatics. https://doi.org/10.1186/s12859-026-06672-0
Image Credits: AI Generated
DOI: 10.1186/s12859-026-06672-0
Keywords: proteomics, tandem mass spectrometry, post-translational modifications, open modification search, spectral library search, peptide identification, modified peptides, fragment-ion shifts, candidate retrieval, proteome informatics, POMS, BMC Bioinformatics
Cite Scienmag News
APA
MLA
Chicago
Kenneth Gardner. (September 23, 2026). New Modification-Aware Search Tool Uncovers Hidden Modified Peptides in Proteomics Data. Scienmag. https://scienmag.com/new-modification-aware-search-tool-uncovers-hidden-modified-peptides-in-proteomics-data/
Kenneth Gardner. “New Modification-Aware Search Tool Uncovers Hidden Modified Peptides in Proteomics Data.” Scienmag, 23 September 2026, https://scienmag.com/new-modification-aware-search-tool-uncovers-hidden-modified-peptides-in-proteomics-data/. Accessed 23 September 2026.
Kenneth Gardner. “New Modification-Aware Search Tool Uncovers Hidden Modified Peptides in Proteomics Data.” Scienmag. September 23, 2026. https://scienmag.com/new-modification-aware-search-tool-uncovers-hidden-modified-peptides-in-proteomics-data/
Copy citation
Download RIS
Tags: BMC Bioinformaticscandidate retrievalcomputational proteomics toolsefficient proteomics workflowsfragment-ion shiftslarge-scale proteomics data analysismass spectrometry data interpretationmodification-aware spectral matchingmodified peptide identificationmodified peptidesopen modification searchopen spectral library search methodspeptide identificationpeptide modification detectionpeptide sequencing accuracyPOMSpost-translational modificationsproteome informaticsProteomicsProteomics mass spectrometryspectral library searchspectral library searchingtandem mass spectrometry
