Fairy: fast approximate coverage for multi-sample metagenomic binning
Jim Shaw, Yun William Yu
Abstract
Abstract Background Metagenomic binning, the clustering of assembled contigs that belong to the same genome, is a crucial step for recovering metagenome-assembled genomes (MAGs). Contigs are linked by exploiting consistent signatures along a genome, such as read coverage patterns. Using coverage from multiple samples leads to higher-quality MAGs; however, standard pipelines require all-to-all read alignments for multiple samples to compute coverage, becoming a key computational bottleneck. Results We present fairy ( https://github.com/bluenote-1577/fairy ), an approximate coverage calculation method for metagenomic binning. Fairy is a fast k-mer-based alignment-free method. For multi-sample binning, fairy can be $$> 250 \times$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mo>></mml:mo> <mml:mn>250</mml:mn> <mml:mo>×</mml:mo> </mml:mrow> </mml:math> faster than read alignment and accurate enough for binning. Fairy is compatible with several existing binners on host and non-host-associated datasets. Using MetaBAT2, fairy recovers $$98.5\%$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mn>98.5</mml:mn> <mml:mo>%</mml:mo> </mml:mrow> </mml:math> of MAGs with $$> 50\%$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mo>></mml:mo> <mml:mn>50</mml:mn> <mml:mo>%</mml:mo> </mml:mrow> </mml:math> completeness and $$< 5\%$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mo><</mml:mo> <mml:mn>5</mml:mn> <mml:mo>%</mml:mo> </mml:mrow> </mml:math> contamination relative to alignment with BWA. Notably, multi-sample binning with fairy is always better than single-sample binning using BWA ( $$> 1.5\times$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mo>></mml:mo> <mml:mn>1.5</mml:mn> <mml:mo>×</mml:mo> </mml:mrow> </mml:math> more $$>50\%$$ <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"> <mml:mrow> <mml:mo>></mml:mo> <mml:mn>50</mml:mn> <mml:mo>%</mml:mo> </mml:mrow> </mml:math> complete MAGs on average) while still being faster. For a public sediment metagenome project, we demonstrate that multi-sample binning recovers higher quality Asgard archaea MAGs than single-sample binning and that fairy’s results are indistinguishable from read alignment. Conclusions Fairy is a new tool for approximately and quickly calculating multi-sample coverage for binning, resolving a computational bottleneck for metagenomics.