Intensity and retention time prediction improves the rescoring of protein‐nucleic acid cross‐links

• DOI: 10.1002/pmic.202300144
• Publication Date: 2024-04-17T14:07:08Z
• Authors: Arslan Siraj; Robbin Bouwmeester; Arthur Declercq; Luisa Welp; Aleksandar Chernev; Alexander Wulf; Henning Urlaub; Lennart Martens; Sven Degroeve; Oliver Kohlbacher; Timo Sachsenberg
• Keywords: 0301 basic medicine; SEQUENCE DATABASES; (MSPIP)-P-2; PEPTIDE IDENTIFICATION; transfer learning; SHOTGUN PROTEOMICS; Mass Spectrometry; STATISTICAL CHARACTERIZATION; 03 medical and health sciences; FALSE DISCOVERY RATES; Nucleic Acids; Medicine and Health Sciences; MS/MS; protein-RNA cross-linking mass spectrometry; Amino Acids; rescoring; fragment peak intensities; Biology and Life Sciences; TANDEM MASS-SPECTRA; 004; SPECTROMETRY; retention time; LIBRARY; RNA; Peptides

In protein-RNA cross-linking mass spectrometry, UV or chemical introduces stable bonds between amino acids and nucleic in complexes that are then analyzed detected spectra. This analytical tool delivers valuable information about RNA-protein interactions RNA docking sites proteins, both vitro vivo. The identification of cross-linked peptides with oligonucleotides different length leads to a combinatorial increase search space. We demonstrate the peptide retention time prediction tasks can be transferred task using simple acid composition encoding, yielding improved rates when error is included rescoring. For more challenging including fragment intensity rescoring, we obtain, on average, similar improvement. Further improvement encoding fine-tuning models might lead further gains, merit research.