RT Journal Article
SR Electronic
T1 Clinical metagenomics bioinformatics pipeline for the identification of hospital-acquired pneumonia pathogens antibiotic resistance genes from bronchoalveolar lavage samples
JF bioRxiv
FD Cold Spring Harbor Laboratory
SP 2020.02.26.966309
DO 10.1101/2020.02.26.966309
A1 Maud Tournoud
A1 Etienne Ruppé
A1 Guillaume Perrin
A1 Stéphane Schicklin
A1 Ghislaine Guigon
A1 Pierre Mahé
A1 Vladimir Lazarevic
A1 Sébastien Hauser
A1 Caroline Mirande
A1 Albrice Levrat
A1 Karen Louis
A1 Gaspard Gervasi
A1 Jacques Schrenzel
YR 2020
UL http://biorxiv.org/content/early/2020/02/27/2020.02.26.966309.abstract
AB Background Shortening the time-to-result for pathogen detection and identification and antibiotic susceptibility testing for patients with Hospital-Acquired and Ventilator-Associated pneumonia (HAP-VAP) is of great interest. For this purpose, clinical metagenomics is a promising non-hypothesis driven alternative to traditional culture-based solutions: when mature, it would allow direct sequencing all microbial genomes present in a BronchoAlveolar Lavage (BAL) sample with the purpose of simultaneously identifying pathogens and Antibiotic Resistance Genes (ARG). In this study, we describe a new bioinformatics method to detect pathogens and their ARG with good accuracy, both in mono- and polymicrobial samples.Methods The standard approach (hereafter called TBo), that consists in taxonomic binning of metagenomic reads followed by an assembly step, suffers from lack of sensitivity for ARG detection. Thus, we propose a new bioinformatics approach (called TBwDM) with both models and databases optimized for HAP-VAP, that performs reads mapping against ARG reference database in parallel to taxonomic binning, and joint reads assembly.Results In in-silico simulated monomicrobial samples, the recall for ARG detection increased from 51% with TBo to 97.3% with TBwDM; in simulated polymicrobial infections, it increased from 41.8% to 82%. In real sequenced BAL samples (mono and polymicrobial), detected pathogens were also confirmed by traditional culture approaches. Moreover, both recall and precision for ARG detection were higher with TBwDM than with TBo (35 points difference for recall, and 7 points difference for precision).Conclusions We present a new bioinformatics pipeline to identify pathogens and ARG in BAL samples from patients with HAP-VAP, with higher sensitivity for ARG recovery than standard approaches and the ability to link ARG to their host pathogens.HAP-VAPHospital-Acquired and Ventilator-Associated PneumoniaDNADesoxyribose Nucleic AcidBALBronchoAlveolar LavageARGAntibiotic Resistance GeneRDBReference DataBaseTBwDMTaxonomic Binning with Determinant MappingTBoTaxonomic Binning only