CheckM

CheckM v2 has been released

CheckM v2 has been released and can be found here. CheckM v1 (this repository) estimates genome completeness and contamination based on the presence or absence of marker genes, i.e., genes that are typically ubiquitous and single copy. In contrast, CheckM v2 uses machine learning models to estimate completeness and contamination. Our assessment indicates CheckM v2 will be more accurate in general. However, you may consider running both CheckM v1 and v2 in some cases since these are independent methodologies for estimating genome quality. It can also be insightful to explore the marker genes that are missing in your genomes, especially if you have multiple related genomes and are interested in systematic loss of an otherwise ubiquitious gene.

Outside of minor updates to fix compatibility issues with external dependencies no maintanence is intended for CheckM v1.

CheckM v1.3 changes

CheckM v1.3 gives the same completeness and contamination estimates as v1.2.8 to v1.2.10, but is faster and uses less memory and disk space. Bug fixes in v1.2.8 slightly changed the estimates of a few genomes (7 of 1,128 test genomes) compared with v1.2.7 and earlier; see Quality estimate changes. CheckM v1.3 also changes some arguments and outputs:

• gPlace replaces pplacer: genomes are placed in the reference tree with gPlace, which gives the same placements as pplacer. pplacer and guppy are no longer required.
• Performance: on 1,000 genomes with 64 threads, lineage_wf takes 19% less time and needs 61% less memory (47 to 18 GiB) and 72% less disk space (4.4 to 1.2 GB). See the performance results.
• Results file: lineage_wf and taxonomy_wf write their results to quality_report.tsv (the qa -o 2 --tab_table table) in the output directory, instead of printing a table.
• Compressed output: most output files are gzip-compressed (e.g., lineage.ms.gz). CheckM still reads the outputs of earlier versions.
• Removed arguments: --reduced_tree and --pplacer_threads (tree, lineage_wf), and -f/--file and --tab_table (lineage_wf, taxonomy_wf).
• New arguments: --gplace_max_memory limits the memory used by gPlace, and --rm_tmp_files removes intermediate files once a workflow has finished.
• CheckM v1.3 requires Python >= 3.11.

Installing and using CheckM

Please see the CheckM documentation for usage details and installation instructions: https://ecogenomics.github.io/CheckM

We do not recommend installing CheckM from the master branch. This may be unstable. Please install an official release of CheckM or use pip.

Dependencies

CheckM requires HMMER (>=3.1), Prodigal (2.6.3) and, from CheckM 1.3.0, gPlace (>=0.1.0) to be on your system path. gPlace replaces pplacer, which was required by earlier versions of CheckM: it gives the same placements as pplacer, bit for bit, using less memory and time. gPlace requires Python >=3.12 and can be installed with pip install gtdb-gplace; it can be installed in a separate environment from CheckM as long as the gplace command is on your system path.

CheckM >= 1.3.0 requires Python >= 3.11 and xopen (>= 2.0), which is installed with CheckM. From CheckM 1.3.0, most output files are gzip-compressed (e.g., lineage.ms.gz, bins/<genome_id>/genes.faa.gz and storage/bin_stats_ext.tsv.gz); CheckM commands still accept the uncompressed output of earlier versions and the uncompressed file names (e.g., checkm qa ./output/lineage.ms ./output).

Estimating quality of CPR genomes

Information about obtaining improved quality estimates for CPR (Patescibacteria) genomes can be found here: https://ecogenomics.github.io/CheckM/workflows.html#using-the-cpr-marker-set

Bug Reports

Please report bugs through the GitHub issues system.

Copyright © 2014 Donovan Parks, Connor Skennerton, Michael Imelfort. See LICENSE for further details.