Accurate Estimation of False Discovery Rates for Protein and Proteoform Identification in Top Down Proteomics

Abstract

Within the last five years, top down proteomics (TDP) has emerged as a high throughput technique for protein identification in addition to characterization and quantitation of thousands of modified proteoforms. Here, a framework for calculating an accurate false discovery rate (FDR) that considers both protein and proteoform levels was used to evaluate local dependencies when aggregating results from replicate LC-MS/MS runs searched with different search modes and parameters.Ê We find that proteoform identifications are not statistically independent of each other and that correcting the FDR locally within a given LC-MS/MS run is not sufficient to control FDR globally across a large experiment.Ê A series of corrections used previously in genomics was implemented to address these issues and produce a global FDR calculation that scales well. ÊThe validity of the system is assessed by analyzing two previously published experimental datasets.ÊWeb-based access via a new TDPORTAL to high-performance computation enables all steps necessary to create a set of results utilizing the accurate and scalable FDR estimation described here. Also, a new application called TOP DOWN VIEWER enables viewing, analyzing, and sharing result sets via .tdReport files and is available at http://topdownviewer.northwestern.edu.