Add to Favourites
To login click here
This article discusses the construction and performance evaluation of a deep learning model for phosphopeptide MS/MS prediction. The model consists of four components: input, the embedding layer, the BERT encoder and the output layer. The model was trained with fourteen higher-energy collisional dissociation (HCD) phosphoproteome datasets, containing >467,000 peptide-spectrum matches (PSMs) of >120,000 mono-phosphorylated peptides, >63,000 multi-phosphorylated peptides and >165,000 non-phosphorylated peptides. The model was trained using the BERT-based model, which converged faster during training for the MS/MS spectra prediction task.
