Version history

Better support for mixed communities and mixed data types.

Several ORF callers, including MetaEuk for splice-aware calling in eukaryotes, can now run in one go, together with your own ORFs. Their calls are merged into loci and clustered into proteins, giving one count table across all sources. With more than one ORF source, only the protein-cluster representatives are annotated, so each protein is annotated once rather than once per caller.

Other highlights:

  • dbCAN CAZyme annotation.
  • QUAST replaces TransRate for assembly statistics.
  • MetaEuk and TransDecoder run on batches of contigs, to limit memory use.
  • Parquet copies of the summary tables with --save_parquet.
  • KofamScan and eggNOG database downloads no longer need network access on the compute node.
  • A new large datasets page, with guidance on assembler memory, resuming a killed Megahit run, and resources for protein consolidation.
  • Fixes for featureCounts on assemblies with more than about 75 million contigs, and for TransDecoder on -resume.

This is a major release because some outputs and parameters change; see Breaking changes in the CHANGELOG. In short: featurecounts/ file names include the ORF caller, columns in the overall_stats tables are renamed (read them by name), samtools idxstats files are only published with --save_idxstats, --skip_qc is gone, and --kofam_dir is now --kofam_dbpath (the old name still works, with a warning).

Thanks to @danilodileo for code and many reviews, to @ewels and @mashehu for code, and to @vagkaratzas, @LuisHeinzlmeier, @piplus2, @Joon-Klaps, @nschan, @leapicard, @tracelail, @jonasscheid, @jfy133, @maxulysse and @pinin4fjords for reviewing.

This patch release fixes BBNorm failing under Singularity/Apptainer when the host’s $TMPDIR isn’t bind-mounted into the container – surfaced as a misleading Cannot run program "bgzip" error even though bgzip is present and working fine inside the container. Reported in #513.

Thanks to @Joon-Klaps , @maxulysse and @FernandoDuarteF for reviewing.

This release addresses key issues with database files:

  • The inbuilt domain for the eggNOG database has disappeared. The download tool now bypasses that to retrieve files from an alternative source.
  • Both the eggNOG and KofamScan databases previously had problems so that previously downloaded files were not recognized because of the lack of a versions.yml file. This should now be fixed.

A number of other smaller bug fixes and additions are included, plus an upgrade to the latest nf-core template and modules.

  • Write fasta files for hmmsearch hits
  • Create Kofamscan output with unique hits
  • Bug fixes
  • nf-test tests for the pipeline

This release addresses a problem with the BBNorm process that sometimes got killed because it used too much memory. In addition, some tools were upgraded, among them all BBMap-based tools.

What’s Changed

New Contributors

Full Changelog: https://github.com/nf-core/metatdenovo/compare/1.0.1...1.1.0

This release fixes two important bugs in 1.0.0:

  • The --eukulele_db parameter did not work and gave syntax error. Besides fixing this issue, it is now not possible to use more than one database per run.
  • User provided assemblies specified with --assembly that were not gzipped were overwritten. This should now be fixed.

Release of nf-core/metatdenovo version 1.0.0