Amit Yadav - Learning from Decoys to Improve the Sensitivity and Specificity of Proteomics Database Search Results

Version 1

      Publication Details (including relevant citation   information):

        Amit Kumar Yadav, Dhirendra Kumar, Debasis Dash (2012) Learning   from Decoys to Improve the Sensitivity and Specificity of   Proteomics Database Search Results. PLoS ONE 7(11): e50651.   doi:10.1371/journal.pone.0050651

      Abstract:

      The statistical validation of database search results is a   complex issue in bottom-up proteomics. The correct and incorrect   peptide spectrum match (PSM) scores overlap significantly, making   an accurate assessment of true peptide matches challenging. Since   the complete separation between the true and false hits is   practically never achieved, there is need for better methods and   rescoring algorithms to improve upon the primary database search   results. Here we describe the calibration and False Discovery   Rate (FDR) estimation of database search scores through a dynamic   FDR calculation method, FlexiFDR, which increases both the   sensitivity and specificity of search results. Modelling a simple   linear regression on the decoy hits for different charge states,   the method maximized the number of true positives and reduced the   number of false negatives in several standard datasets of varying   complexity (18-mix, 49-mix, 200-mix) and few complex datasets (E.   coli and Yeast) obtained from a wide variety of MS platforms. The   net positive gain for correct spectral and peptide   identifications was up to 14.81% and 6.2% respectively. The   approach is applicable to different search methodologies-   separate as well as concatenated database search, high mass   accuracy, and semi-tryptic and modification searches. FlexiFDR   was also applied to Mascot results and showed better performance   than before. We have shown that appropriate threshold learnt from   decoys, can be very effective in improving the database search   results. FlexiFDR adapts itself to different instruments, data   types and MS platforms. It learns from the decoy hits and sets a   flexible threshold that automatically aligns itself to the   underlying variables of data quality and size.

      Address (URL): http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0050651