ARGminer: a web platform for the crowdsourcing-based curation of antibiotic resistance genes

• DOI: 10.1093/bioinformatics/btaa095
• Publication Date: 2020-02-08T12:10:08Z
• Authors: G A Arango-Argoty; G K P Guron; E Garner; M V Riquelme; L S Heath; A Pruden; P J Vikesland; L Zhang
• Keywords: Crowdsourcing; Web resource

Curation of antibiotic resistance gene (ARG) databases is a labor-intensive process that requires expert knowledge to manually collect, correct, and/or annotate individual genes. Correspondingly, updates existing tend be infrequent, commonly requiring years for completion and often containing inconsistences. Further, because limitations manual curation, most ARG contain only small proportion known ARGs (~5k genes). A new approach needed achieve truly comprehensive database, while also maintaining high level accuracy. Here we propose web-based curation system, ARG-miner, which supports annotation at multiple levels, including: name, category, mechanism, evidence mobility occurrence in clinically-important bacterial strains. To overcome employ crowdsourcing as novel strategy expanding capacity towards achieving comprehensive, up-to-date database. We develop validate the by comparing performance cohorts curators with varying levels expertise, demonstrating ARG-miner more cost effective less time-consuming relative traditional curation. further demonstrate reliability trust validation filter rejecting confounding input generated spammers. Crowdsourcing was found accurate annotation, an accuracy >90% diverse test set ARGs. provides public API database available http://bench.cs.vt.edu/argminer.