Description

Microbial genome-wide association studies (GWAS) have the potential to illuminate complex genotype-phenotype associations in bacterial populations. Efficient computational methods applied to \textit{de novo} assembled contigs provide insights into the genetic causes of pathogenic traits of interest such as virulence and antibiotic resistance. PSL-GWAS is a novel microbial GWAS method that finds associations using Probabilistic Soft Logic, a probabilistic programming framework based on statistical relational machine learning. PSL-GWAS implicitly models population structure and can jointly infer over multiple phenotypes concurrently.