Table 2. Genes interacting with four drugs in the Johnson et al (2019) dataset (with outlier negative slope in CGA-LMM model and Zrobust<-3.5).

drug	Significantly interacting genes (slope<0, Zrobust<-3.5) (rank # out of 152 genes in hypomorph library)	Zrobust
trimethoprim	#1: lipA—lipoyl transferase	-9.6
	#2: trpG—aminodeoxychroismate synthase	-9.5
	#3: desA1—fatty acid desaturase	-4.4
	#4: nusA–transcription termination/antiterm. protein	-3.8
methotrexate	#1: trpG—aminodeoxychroismate synthase	-6.0
	#2: gyrA—DNA gyrase	-4.8
	#3: ftsK—ATPase involved in cell division	-4.2
	#4: gca—GDP-mannose dehydratase	-3.0
rifampin	#1: dapF—diaminopimelate epimerase	-6.4
	#2: Rv3267	-6.1
	#3: desA1—fatty acid desaturase	-4.3
	#4: acn—aconitase	-3.9
	. . .	---
	#14: rpoB—RNA polymerase beta subunit	-1.7 (n.s)
BRD-4592	#1: desA1—fatty acid desaturase	-5.0
	#2: ftsK—ATPase involved in cell division	-4.3
	#3: Rv3267	-4.1
	#4: dapF—diaminopimelate epimerase	-3.9

Out of a library of 155 genes, there were 4 genes with outlier negative slopes for each drug. Genes relevant to the mechanism of resistance are bold-faced. Note that rpoB was ranked highly for rifampin, though it did not exceed the Zrobust cutoff (n.s. = not significant). Also, the expected target of BRD-4592, trpA, was not represented in the hypomorph library.