Abstract
Advances in genome engineering have enabled routine engineering and interrogation of microbial resistance on a scale previously impossible, but developing an integrated understanding of resistance from these data remains challenging. As part of our continued efforts to address this challenge, we present a significant update of our previously released Resistome database of standardized genotype-resistance phenotype relationships, along with a new web interface to enable facile searches of genomic, transcriptomic, and phenotypic data within the database. Revisiting our previous analysis of resistance, we again find distinct mutational biases associated with random selection versus genome-scale libraries, along with pervasive pleiotropy among resistant mutants. Attempts to predict mutant phenotypes using machine learning identified the lack of comprehensive phenotype screening and small size of the Resistome corpus as challenges for effective model training. Overall, the Resistome represents a unique platform for understanding the interconnections between both current and future resistant mutants, and is available for use at https://resistome-web-interface.herokuapp.com.
