PepBindPred is for users interested in identifying peptide binding regions within a protein sequence, which may bind to a know protein structure. PepBindPred was trained on a set of non-redundant protein sequences containing known peptide binding regions sourced from the ELM database.
Input:This tool takes as input a UniProt sequence ID. This is the sequence in which you wish to identify peptide binding regions. At present PepBindPred predicts peptide binding regions in 50 residue windows so therefore you must provide the number of the residue within this sequence where you would like the search to start.
PepBindPred also takes as input the PDB ID of a structure that you wish to predict binding with. The chain id of a bound peptide must be provided as we use this to define the search space used by the Vina docking program.
The method uses the entire protein sequence for the prediction of secondary structure by Distill and disorder by IUpred, which are provided as additional inputs to the neural network along with the Vina docking scores. The maximum length of a protein sequence that can be analysed is 10,000 residues.
Output:The tool outputs a plot of the PepBindPred score for requested the 50 residue window. This score represents the tool's prediction of the likelihood that each residue is part of a peptide binding region within that protein sequence. Information about the Relative Local Conservation of each residue and the disorder context of the protein are included for the whole protein sequence.