P181-S A New Software Tool to Facilitate False-Positive Rate Assessment in Protein Identification

There are currently many employed variants of a class of methods for assessing the false-positive rates in protein identification results, referred to generically here as “decoy database searching.” The general idea of all variants is to assess false-positive rates by presenting identification software with some component of potential answers that are known to be incorrect. The major variants in this class of methods differ primarily in (1) the type of decoy component (reversed proteins, randomized sequences, or other more sophisticated approaches) and (2) whether the valid sequences and decoy component are searched separately or together competitively. Regardless of which specific approach is preferred, it is clear that the growing popularity of this class of methods stems largely from the ability of these approaches to provide an assessment of the accuracy of identification results that is virtually independent of the statistical approach used by the software that generated the result. This allows rigorous comparison of results derived from different software tools, as well as different instruments, sample preparations, and separation methods. We present here a software tool that is designed to facilitate result quality assessment by decoy methods. The tool provides an easy distillation of the yield of true positives at critical values of false-positive rates for both peptides and proteins. It also provides this information using two different types of false-positive rate calculations—the commonly used approach of assessing the error rate for all detections over a threshold, referred to as the “aggregate” rate, and also the “instantaneous” false-positive rate, which indicates the chance of error for detections with similar scores. The tool can be operated in a standalone mode to analyze the output from many search engines, or it can be operated as integrated into ProteinPilot software.