DbyDeep: Exploration of MS-Detectable Peptides via Deep Learning

Abstract

Predicting peptide detectability is useful in a varietyof massspectrometry (MS)-based proteomics applications, particularly targetedproteomics. However, most machine learning-based computational methodshave relied solely on information from the peptide itself, such asits amino acid sequences or physicochemical properties, despite thefact that peptides detected by MS are dependent on many factors, includingprotein sample preparation, digestion, separation, ionization, andprecursor selection during MS experiments. DbyDeep (Detectabilityby Deep learning) is an innovative end-to-end LSTM network model forpeptide detectability prediction that incorporates sequence contextsof peptides and their cleavage sites (by protease). Utilizing thecleavage site contexts could improve the performance of prediction,and DbyDeep outperformed existing methods in predicting peptides recognizablefrom multiple MS/MS data sets with diverse species and MS instruments.We argue for the necessity of a learning model that encompasses severalcontexts associated with peptide detection, as opposed to dependingjust on peptide sequences. There is a Python implementation of DbyDeepat https://github.com/BISCodeRepo/DbyDeep.