Seminarios Institucionales

Seminario 05/07: How clear is our current view on microbial dark-matter? (Re-)assessing public MAG & SAG-datasets with "MDMcleaner"

Título: How clear is our current view on microbial dark-matter? (Re-)assessing public MAG & SAG-datasets with "MDMcleaner"

Disertante: John Vollmers, Karlsruhe Institute of Technology


As of today, the majority of environmental microorganisms, designated as "microbial dark matter" (MDM), still remains uncultured. Insights into this uncultured majority are therefore predominantly limited to genome reconstructions via cultivation independent approaches such as single cell amplified genomes (SAGs) and metagenome assembled genomes (MAGs). Both, however, are sensitive to contaminations, potentially causing misattributed contigs to be included in the respective reconstructed genomes. Such contaminations may bias comparative genome analyses and cause false conclusions.
Consequently, strict contamination filtering needs to be applied. Current genome reporting standards, however, emphasize completeness over purity. Furthermore, current de facto standard genome assessment tools are most biased in the case of uncultured taxa and fragmented genomes, thereby discriminating against many SAGs and MAGs. The result is a potential gradual reference database corruption that may cause further error propagations and increasingly distort our view on microbial dark matter.
To combat this, we present a python implementation of a novel contig classification, screening and filtering workflow that tackles these issues: "MDMcleaner". This workflow reports potential contaminants not only in the subject genomes but also in the underlying reference datasets, allowing to simultaneously expand and refine genome databases.
Subjected to current "high quality" genome datasets, MDMcleaner revealed substantial fractions of preventable contaminations overlooked by current screening approaches.
Affected genomes included selected "representative genomes" of curated public reference datasets. By sensitively detecting potential contaminants and eliminating error propagation from contaminated references, MDMcleaner can substantially enhance our view on "microbial dark matter" genomics.