The emergence and prevalence of drug resistance is an inevitable consequence of the biological response to selective pressures like those that accompany drug therapies. We therefore developed an unbiased and systematic screen to identify the functional impact of drug resistant variants in a fast and cost-effective manner. Current methods to understand and predict drug resistance often require significant time, computational power and clinical constraints. These challenges are exacerbated by the lack of statistical methods to accurately call allele frequency variants.
By combining deep sequencing and a Bayesian statistical model we provide a comprehensive survey of drug resistance alleles from complex variant populations. Using dihydrofolate reductase (DFR1), the target of methotrexate chemotherapy as a benchmark we identified and validated functional alleles correlated with methotrexate (MTX) resistance in yeast. This massively parallel strategy allowed us to catalog alleles with dominant resistance phenotypes independent of pre-existing competitive fitness traits. We confirmed 10 MTX-resistant missense mutations in DFR1/dfr1Δ heterozygotes, 2 of which also confer growth advantages and 3 hypomorphic alleles. In parallel we confirmed 8 functional missense mutations in the haploid dfr1Δ strain. The recovery of these distinct dfr1 MTX resistant-alleles underlines the power of using genetic-based assay in S. cerevisiae to dissect complex functional determinants of drug resistance. Because many of these novel resistant variants map to structurally conserved binding pockets within the Dfr1 active site, this approach should yield insight into the mechanistic differences between antifolate resistant alleles and folate metabolism.