Source-linked AI summary
CoverM: Read alignment statistics for metagenomics
Samuel T. N. Aroney, Rhys J. P. Newell, Jakob N. Nissen, Antonio Pedro Camargo, Gene W. Tyson, Ben J. Woodcroft
TL;DR
Metagenomic coverage calculations differ across tools despite supporting both genome recovery and community profiling. CoverM unifies contig- and genome-level coverage statistics using Mosdepth arrays and filtered read alignments, while its relative-abundance approach assumes comparable genome sizes and sufficiently complete, uncontaminated references.
Problem
Coverage statistics are central to differential-coverage binning and community profiling, but software packages use substantially different definitions and algorithms.
Method
CoverM aligns reads to contig or genome references, filters alignments, processes them with Mosdepth arrays, and aggregates coverage statistics at contig or genome level.
Results
CoverM provides efficient calculation of several coverage metrics for contig-based binning and genome-based community profiling.
Takeaways & Limitations
A single flexible tool supports coverage-based metagenomic binning and community profiling across contig and genome references.
Takeaways & Limitations
The relative-abundance formula assumes similar average genome sizes between referenced and missing community members and roughly complete, uncontaminated references.
Abstract
from arXiv · showhide
Genome-centric analysis of metagenomic samples is a powerful method for understanding the function of microbial communities. Calculating read coverage is a central part of analysis, enabling differential coverage binning for recovery of genomes and estimation of microbial community composition. Coverage is determined by processing read alignments to reference sequences of either contigs or genomes. Per-reference coverage is typically calculated in an ad-hoc manner, with each software package providing its own implementation and specific definition of coverage. Here we present a unified software package CoverM which calculates several coverage statistics for contigs and genomes in an ergonomic and flexible manner. It uses 'Mosdepth arrays' for computational efficiency and avoids unnecessary I/O overhead by calculating coverage statistics from streamed read alignment results. CoverM is free software available at https://github.com/wwood/coverm. CoverM is implemented in Rust, with Python (https://github.com/apcamargo/pycoverm) and Julia (https://github.com/JuliaBinaryWrappers/CoverM_jll.jl) interfaces.
Introduction
Genome-centric metagenomics relies on read coverage for genome recovery and community profiling, but coverage calculations vary substantially across tools. CoverM addresses this fragmentation by calculating multiple contig- and genome-level statistics in one tool.
- Genome-centric metagenomics can recover complete or near-complete genomes from deeply sequenced microbial-community DNA.
- Differential coverage binning groups contigs with similar abundances across samples to improve sequence-agnostic genome recovery.
- Coverage statistics derived from BAM alignments vary substantially among metagenomic binning tools.
- Mosdepth arrays record changes in mapped-segment counts by position and were shown to be two times faster than naive exact mean-coverage methods.
- CoverM uses Mosdepth arrays to efficiently calculate multiple coverage statistics for both contigs and genomes.
Software Implementation
CoverM accepts reads or BAM files and contig or genome references, processes filtered alignments with Mosdepth arrays, and reports contig- or genome-level coverage. Its genome profiling includes relative abundance scaling and a default coverage threshold.
- CoverM calculates coverage statistics for contigs or genomes from metagenomic or microbial-isolate reads.
- In contig mode, CoverM aligns reads to contigs and reports contig statistics; in genome mode, it aggregates contig coverage and reports genome statistics.
- CoverM accepts reads or precalculated BAM files with contig or genome references, and can dereplicate genomes before alignment.
- Reads are aligned with Strobealign, Minimap2, or BWA, filtered by alignment quality criteria, and processed using Mosdepth arrays.
- The default genome statistic is relative abundance, with unaligned-read fractions used to scale abundances when the reference database is incomplete.
- Equation 1 assumes equal average genome sizes for referenced and missing community members and roughly complete, uncontaminated reference genomes.
Conclusion
CoverM provides efficient, flexible coverage calculations for contig-based metagenomic binning and genome-based community profiling.
- CoverM calculates several coverage metrics for contig-based binning and genome-based community profiling.The implementation offers more than 50 arguments and options for input/output, alignment, filtering, dereplication, and coverage metrics.
Funding
The work received support from multiple research funding organizations and grants.
- The work was supported by EMERGE, the U.S. Department of Energy, the Australian Research Council, and FAPESP.