An improved statistical method to identify chemical-genetic interactions by exploiting concentration-dependence

• DOI: 10.1371/journal.pone.0257911
• Publication Date: 2021-10-01T18:07:12Z
• Authors: Esha Dutta; Michael A. DeJesus; Nadine Ruecker; Anisha Zaveri; Eun-Ik Koh; Christopher M. Sa...; Dirk Schnappinger; Thomas R. Ioerger
• Keywords: 0303 health sciences; Science; Q; R; 612; Mycobacterium tuberculosis; Peptidoglycans; Drug interactions; Polymerase chain reaction; Anti-Bacterial Agents; 03 medical and health sciences; Cholesterol; Methotrexate; Antibiotics; Genes, Bacterial; Drug Discovery; Isoniazid; Medicine; Research Article

Chemical-genetics (C-G) experiments can be used to identify interactions between inhibitory compounds and bacterial genes, potentially revealing the targets of drugs, or other functionally interacting genes pathways. C-G involve constructing a library hypomorphic strains with essential that knocked-down, treating it an compound, using high-throughput sequencing quantify changes in relative abundance individual mutants. The hypothesis is that, if target drug same pathway are present library, such will display excessive fitness defect due synergy dual stresses protein depletion antibiotic exposure. While assays at single concentration susceptible noise yield false-positive interactions, improved detection achieved by requiring gene concentration-dependent. We novel statistical method based on Linear Mixed Models, called CGA-LMM, for analyzing data. approach designed capture dependence each hypomorph increasing concentrations through slope coefficients. To determine which represent candidate CGA-LMM uses conservative population-based negative slopes considered significant only they outliers respect rest population (assuming most do not interact given inhibitor). applied analyze 3 independent libraries M . tuberculosis antibiotics anti-tubercular activity, we known expected 7 out 9 drugs where relevant known.