Figure 1

Schema of the GWAMAR pipeline. For a set of considered bacterial strains, the input data for GWAMAR consists of (i) a set of mutations; (ii) a set of drug resistance profiles; and (iii) phylogenetic tree for the set of bacterial strains. Typically the set of mutation profiles is generated using eCAMBer, which is able to download the genome sequences and annotations for the set of bacterial strains, identify point mutations based on multiple alignments, and reconstruct the phylogenetic tree of the considered bacterial strains. The first step of GWAMAR is to compute binary mutation profiles for all point mutations. This step significantly reduces the number of genetic profiles considered. Finally, GWAMAR implements several statistical scores to associate drug resistance profiles with mutation profiles. These include: mutual information, odds ratio, hypergeometric test, weighted support and tree-generalized hypergeometric score (TGH). As a result, we obtain ordered lists of drug resistance associations, where the top scored associations are the most likely to be real.