Pepfind: Protein Identification Algorithm-based Target-decoy Database Feature Information Matching For Tandem Mass Spectrometry

In the database search methods, it is the core of protein identification algorithmsfor experimental tandem mass spectra(LC-MS/MS) to match with all possiblecandidate peptides in database( Peptides matching score model). Usually, because ofMS/MS with complex noise peak, there is almost no database search algorithm thatcan identify all the sequence of peptides in the sample. Various search matchingalgorithms of peptides scoring function can only adopt different ways to considermore MS/MS characteristics and influence factors to increase the amount ofidentification of peptides. In most matching scoring function of database searchalgorithms, researchers mainly considerate the number of matching peaks, the numberof continuous matching peaks and the characteristics of preset peak intensity. InPro Ver B algorithms, we integrated the characteristics of pair-wise amino acid(beforeand after amino acid of fragment ions) that the amount of identification has beenincreased obviously which compared with Mascot and Sequest. It is shown that thescoring function added more characteristics of MS/MS can significantly improve theamount of peptides identified.In this paper, we presented a novel concept scoring model of peptideidentification algorithms named Pep Find. we statistics the different type of fragmentions matching the target-decoy peptides sequence library which based on various MZerror and intensity range, extracting the feature information of statistical results, thenquantified by mathematical methods, finally, integrated the quantified information toscoring model of Pep Find. This new algorithms is compared with commonly usedcommercial software at 1% FDR. The results show that Pep Find is robust andversatile for various datasets obtained from different instruments. We expect thatPep Find will be broadly applied in the proteomics studies.