Author:

Abstract

Disease-specific protein biomarkers can assist in the disease identification process utilizing protein inference. These biomarkers play a major role in the drug discovery process. Biomarker discovery consists of a set of phases starting from mass spectrum files of peptides or proteins and ending with some significantly expressed proteins of a particular disease condition. Different techniques and tools have been introduced to perform protein inference and biomarker identification, and it still requires improvements in the accuracy of the protein and biomarker identification process. Further, it requires improvements in speed as it consumes hours or days to carry out the processes of protein biomarker discovery. In this paper, we thoroughly present and validate the Open Pipeline for Biomarker Identification (OPBI) on six different datasets and show how the pipeline fits into the process of protein identification with hardware acceleration. OPBI uses the information of tandem mass spectrometry (MS2) and the first stage of mass spectrometry (MS1). It achieved 0.0003–0.0004 false discovery rate and 2–3 times of speed-up with respect to existing MaxQuant software in different contexts on a general purpose computer with Intel Core i7 processor of 3.4 GHz frequency and 12 GB memory. Furthermore, the identified biomarkers can be utilized with the FPGA accelerated protein identification framework. According to the results observed, a considerable speed-up is achieved in the whole process of protein inference as well as peptide matching and peptideprotein mapping process. It further provides a methodology for the downstream analysis of protein biomarkers.

Keywords: Accuracy, biomarker-discovery, efficiency, fold-change-ratio, mass-spectrometry, shotgun-proteomics