The PepCleave algorithm predicts peptides resulting from proteasomal cleavage. For a given set of proteins, it provides all predicted supra-threshold peptides of a given length. The algorithm uses the sequence of the peptide and its two flanking positions. In the absence of flanking positions, the algorithm still works but it is slightly less precise.
The PepCleave algorithm can use five different scores. The different scores were learnt either on different databases or with a different bias for sensitivity and specificity. For a given score, a peptide is defined as cleaved, if it has a positive score, and not cleaved otherwise.
The different score are:
- The most basic score predicts proteasomal cleavage in general. It the basic score to be used and it minimize sensitivity and specificity equally.
- The Constitutive proteasome score - This score predicts cleavage products of the constitutive proteasome.
- The Immunoproteasome score - This score is to be used when cleavage products are predicted in the
presence of IFN-[if gte vml 1]>
- The good over weighted score. This score is more tolerant. It is to be used when one wants to make sure that no produced cleaved peptide is missed. In the case that this score predicts a peptide is not produced, it is very likely that this peptide is indeed not produced. However, if it predicts a peptide that is indeed produced, there is a higher chance it is an erroneous prediction.
- The bad over weighted score. This is a stricter score. This score is meant to be used when one wants to minimize the possibility that a non-cleaved peptide will be predicted as if it should be cleaved. When this score predicts a peptide is produced - it is most likely to be the actual situation.
Input and output format
The input format is Fasta. The output is a list of predicted epitopes and their score.
A full explanation
A full explanation can be found at: Full peptide proteasomal cleavage score (pdf, 161Kb)