(A) A diagram of the ProteinProphet algorithm. The numbers in the circles correspond to the steps mentioned in the text. The presence of loops in the diagram represents iterative inference algorithms used by ProteinProphet. (B) Toy examples illustrating the impact of fluctuations in peptide identification probabilities on the inference outcome from ProteinProphet (version 4.2 RAPTURE rev 2). Top part shows four identified peptides corresponding to two proteins. Peptides p1 and p2 are shared peptides while peptides p3 and p4 are unique peptides for proteins P1 and P2, respectively. The numerical values are the peptide identification probabilities used in the toy examples. Bottom: ProteinProphet results on seven data sets with minor changes in peptide identification probabilities. Noisy Input and Output: the peptide and protein identification probabilities respectively; group: protein group probability; N/A: not reported in ProteinProphet output.