Table I Criteria for Manual PrSM validation
SectionMetricHeuristic
PrecursorNumber of charge statesMore charge states are better
AbundanceHigher is somewhat better
m/z ShiftsCheck for the isotopic envelope shifts, either consistently or randomly
FragmentsNumber of fragmentsMore is better
Location of fragmentsConsecutive fragments and “golden” pairs (e.g. B and Y ions at the same position) carry more weight
Preferred fragmentationCertain fragmentation methods have preferred sites (e.g. d-P for HCD)
% Ion current explainedHigher is better, but not required (co-isolation)
Consistent m/z shiftsMatching error of fragments from a single spectrum should be normally, not uniformly, distributed
DuplicationCheck for ions counting twice (e.g. mass of B40 = mass of Y41 within tolerance)
CandidateNumber of modificationsGenerally, less is better
Types of modificationsThere are tiers of modifications, where some are just more common than others
Other hits, same searchSeeing the same candidate elsewhere lends support
Known homologyIf the candidate belongs to a family with many similar sequences (e.g. histones) ensure that another family member isn't better
Family of modified formsIf one sees a highly modified form, there should also be other modified forms