Benchmarking immunoinformatic tools for the analysis of antibody repertoire sequences

• DOI: 10.1093/bioinformatics/btz845
• Publication Date: 2019-12-20T12:11:15Z
• Authors: Erand Smakaj; Lmar Babrak; Mats Ohlin; Mikhail Shugay; Bryan Briney; Deniz Tosoni; Christopher Galli; Vendi Grobelsek; Igor D’Angelo; Branden Olson; Sai Reddy; Victor Greiff; Johannes Trück; Susanna Marquez; William Lees; Enkelejda Miho
• Keywords: Benchmarking; Complementarity (molecular biology); Benchmark (surveying)

Abstract Summary Antibody repertoires reveal insights into the biology of adaptive immune system and empower diagnostics therapeutics. There are currently multiple tools available for annotation antibody sequences. All downstream analyses such as choosing lead drug candidates depend on correct these sequences; however, a thorough comparison performance has not been investigated. Here, we benchmark commonly used immunoinformatic tools, i.e. IMGT/HighV-QUEST, IgBLAST MiXCR, in terms reproducibility output, accuracy speed using simulated experimental high-throughput sequencing datasets. We analyzed changes IMGT reference germline database last 10 years order to assess output. found that only 73/183 (40%) V, D J human genes were shared between sets by tools. results differed In alignment accuracy, MiXCR had highest average frequency gene mishits, 0.02 mishit lowest, 0.004 frequency. Reproducibility output complementarity determining three regions (CDR3 amino acids) ranged from 4.3% 77.6% with preprocessed data. addition, run time was assessed: fastest tool number sequences processed per unit time. These indicate greatly choice bioinformatics tool. Our support informed decision-making immunoinformaticians based repertoire composition platforms. Availability implementation utilized paper free academic use. Supplementary information data at Bioinformatics online.