Identification and quantitation of clinically relevant microbes in patient samples Comparison of three k-mer based classifiers for speed, accuracy, and sensitivity

摘要

Currently, the gold standard for identifying pathogens that are causing infection is to attempt growth in culture followed by identification based on physical characteristics such as shape and metabolic profile. However, many organisms do not grow in culture or are overgrown by faster growing organisms that out-compete them. Another method to identify pathogens in infections is to sequence the DNA in the samples and use that DNA sequence to identify the pathogens presenta process called metagenomic sequencing. Analyzing clinical metagenomic data can be difficult given the amount of data generated, high levels of human DNA contamination and a lack of well-defined bioinformatics methods. In this study, three leading software tools were compared for identification and quantitation of microbes in metagenomic data. One tool, called Centrifuge, reliably identified microbes present at just 1% relative abundance while requiring less computer time and resources than the others to which it was compared. Moreover, we found that Centrifuge results changed minimally when time-consuming quality control and host-screening steps were eliminated. We also examined Centrifuge's performance in real-word clinical data sets showing that Centrifuge identified the same pathogens as culture.