Prediction of LC-MS/MS Properties of Peptides from Sequence by Deep Learning*[S]
- Resource Type
- Article
- Authors
- Guan, Shenheng; Moran, Michael F.; Ma, Bin
- Source
- Molecular and Cellular Proteomics (MCP Online); October 2019, Vol. 18 Issue: 10 p2099-2107, 9p
- Subject
- Language
- ISSN
- 15359476; 15359484
Indexed retention times (iRT), MS1 (the first level of mass analysis) or survey scan charge state distributions, and sequence ion intensities of MSMS (tandom mass spectrometry) spectra were predicted from peptide sequence by use of long-short term memory (LSTM) recurrent neural networks models. Data points on the order of 105were used to train the iRT and charge state distribution models. An HCD sequence ion prediction model was trained with 2×106experimental spectra. The models with a simple deep learning architecture can predict those three key LC-MS/MS (Liquid chromatography-tandem mass spectrometry) properties with superior accuracies.