Contamination detection in genomic data: more is not enough

Abstract

The decreasing cost of sequencing and concomitant augmentation of publicly available genomes have created an acute need for automated software to assess genomic contamination. During the last 6 years, 18 programs have been published, each with its own strengths and weaknesses. Deciding which tools to use becomes more and more difficult without an understanding of the underlying algorithms. We review these programs, benchmarking six of them, and present their main operating principles. This article is intended to guide researchers in the selection of appropriate tools for specific applications. Finally, we present future challenges in the developing field of contamination detection.

Keywords: Algorithms; Contamination detection; Corroboration; Databases; Genomics; Review.

Fig. 1

Sources of genomic contamination. Three types of issues lead to contamination of genomic sequence data: biological, experimental and computational. The contamination of “pure” cultures can be due to both experimental (e.g. accidental introduction of contaminating microorganisms) and biological causes (e.g. the presence of an endosymbiont). Redundant contamination occurs when a genomic segment is present multiple times in a genome (e.g. multiple SSU rRNAs from different organisms). Non-redundant contamination occurs when a genomic region of the main organism, the expected one, is replaced by the corresponding region of a foreign organism (e.g. the SSU rRNA of the main organism is replaced by the SSU rRNA from a foreign organism). An extra DNA segment, not part of the main organism but belonging to a contaminant, would also be considered as a non-redundant contamination (e.g. eukaryotic DNA in a bacterial genome). A mixed scenario is also possible, as represented in the redundant contamination part of the figure

Fig. 2

Overview of algorithms. The algorithms are clusterized based on their operating principles, as described in the section “Overview of algorithms”. Squares on the top of the figure represent specific features of the algorithms. Non-redundant means that the software can detect contaminant genes without equivalent in the surveyed genome. Intra-species means that the algorithm can detect contamination at the species level. Inter-domain means that the algorithm can detect prokaryotic and eukaryotic contamination simultaneously. Database features show that the algorithm can use the GTDB Taxonomy and/or a moderately contaminated reference database. Expected organism indicates whether the algorithm can detect the main organism by itself and/or if the user can specify it. Additional functionalities list interesting peculiar functions of the programs, such as outputting the completeness of a genome, cleaning a genome from its contaminants, filtering reads based on their taxonomy (positive filtering), or enriching Multiple Sequence Alignments (MSAs) in orthologous sequences while controlling the taxonomy