Abstract
We have developed an easy-to-use and memory-efficient method called PhenotypeSeeker that (a) generates a k-mer-based statistical model for predicting a given phenotype and (b) predicts the phenotype from the sequencing data of a given bacterial isolate. The method was validated on 167 Klebsiella pneumoniae isolates (virulence), 200 Pseudomonas aeruginosa isolates (ciprofloxacin resistance) and 460 Clostridium difficile isolates (azithromycin resistance). The phenotype prediction models trained from these datasets performed with 88% accuracy on the K. pneumoniae test set, 88% on the P. aeruginosa test set and 96.5% on the C. difficile test set. Prediction accuracy was the same for assembled sequences and raw sequencing data; however, building the model from assembled genomes is significantly faster. On these datasets, the model building on a mid-range Linux server takes approximately 3 to 5 hours per phenotype if assembled genomes are used and 10 hours per phenotype if raw sequencing data are used. The phenotype prediction from assembled genomes takes less than one second per isolate. Thus, PhenotypeSeeker should be well-suited for predicting phenotypes from large sequencing datasets.
PhenotypeSeeker is implemented in Python programming language, is open-source software and is available at GitHub (https://github.com/bioinfo-ut/PhenotypeSeeker/).
Summary Predicting phenotypic properties of bacterial isolates from their genomic sequences has numerous potential applications. A good example would be prediction of antimicrobial resistance and virulence phenotypes for use in medical diagnostics. We have developed a method that is able to predict phenotypes of interest from the genomic sequence of the isolate within seconds. The method uses statistical model that can be trained automatically on isolates with known phenotype. The method is implemented in Python programming language and can be run on low-end Linux server and/or on laptop computers.