Statistical assessment for mass-spec protein identification using peptide fingerprinting approach

We derive and validate a novel statistical model for confidence assessment of protein identification results using peptide mass fingerprint data. We simulate the digestion of the proteins and compare each peptide mass with the input mass. We compute scores from this matching of peptide and compute the distribution of scores for all the proteins in the database. Based on the distribution, we can provide the expectation value for a protein match in the database. We conclude that, given the complexity and noise of the data, the best method for effective confidence matching is using one scoring scheme for matching and another scoring scheme for confidence assessment.