Whole proteome-level GPS predictions (part 2)

This is the expanded set of all predictions for GPS, run on the entire reference proteome, including sites not known to be phosphorylated. This dataset is used to perform a fast update when new phosphosites are discovered. The uncompressed folder will yield a large CSV file with predictions in list format (i.e. one line per kinase-substrate prediction) Columns in this order:substrate_id - unique substrate (accession_site) IDsubstrate_acc - Uniprot accession of substrate proteinsubstrate_name - Name of proteinsite - amino acid type and position (S5, means serine position 5)pep - 15-amino acid sequence centered on site of phosphorylationscore - prediction algorithm scoreKinase Name - name of kinase by our controlled ontology (found in this project) Each entry indicates the protein at position (identified by peptide) and has a score weight prediction for the given kinase. FOR FULL GPS RAW, you must combine this with the first zip part. Please be sure to download both into the same directory before unzipping. The final, uncompressed file, is 24GB.