文摘
This paper investigates the use of survival functions andexpectation values to evaluate the results of proteinidentification experiments. These functions are standardstatistical measures that can be used to reduce variousprotein identification scoring schemes to a common,easily interpretably representation. The relative merits ofscoring systems were explored using this approach, aswell as the effects of altering primary identification parameters. We would advocate the widespread use of thesesimple statistical measures to simplify and standardizethe reporting of the confidence of protein identificationresults, allowing the users of different identificationalgorithms to compare their results in a straightforwardand statistically significant manner. A method is describedfor measuring these distributions using information thatis being discarded by most protein identification searchengines, resulting in accurate survival functions that arespecific to any combination of scoring algorithms, sequence databases, and mass spectra.