Comparative Analysis of Data-Driven Rescoring Platforms for Improved Peptide Identification in HeLa Digest Samples

Abstract

Mass spectrometry is a critical tool to understand complex changes in biological processes. Despite significant advances in search engine technology, many spectra remain unassigned. This research evaluates the performance of three rescoring platforms, Oktoberfest, MS²Rescore, and inSPIRE, using MaxQuant output. The results indicated a substantial increase in identifications at the peptide level (40%–53%) and PSM level (64%–67%). However, some peptides were lost due to limitations in processing posttranslational modifications (PTMs)—with up to 75% of lost peptides exhibiting PTMs. Each platform displayed distinct strengths and weaknesses. For instance, inSPIRE performed best in terms of peptide identifications and unique peptides, while MS²Rescore performed better for PSMs at higher FDR values. Differences in platform performance stemmed from different sources: original search engine feature selection, type of ion series predicted, retention time predictor, and PTMs compatibility. Overall, inSPIRE showed a superior ability to harness original search engine results. Taken all together, rescoring platforms clearly outperformed original search results; however, they demanded additional computation time (up to 77%) and manual adjustments. The findings here underline the necessity of integrating rescoring platforms into current proteomics pipelines but also address some challenges in their implementation and optimization. Future integrated platforms may help enhance adoption.