Added

• #182 - Added an alternative, iterative family generation algorithm, selectable with --family_generation_algorithm iterative. It hands whole chunks of clusters (--clusters_per_chunk) to mgnifam, which repeats HMM building, sequence recruitment and realignment per cluster until each family converges or is discarded. Applies to both newly created and merged families. (by @vagkaratzas)

Fixed

• #182 - Fixed MERGE_FAMILIES:MERGE_SEEDS using the first sample’s seed MSAs for every sample: pooled families are now paired with their own sample’s seeds. (by @vagkaratzas)
• #177
  • Fixed OSError: [Errno 36] File name too long in MERGE_FAMILIES:MERGE_SEEDS when a pool combines many families: merged_id now collapses to a stable hash once the readable form would exceed the filesystem name-length limit. (by @vagkaratzas, bug reported by @Yixuan39)
  • Fixed FAMSA_ALIGN aborting with exit status 134 on merged families: MERGE_SEEDS now drops all-gap records, which CLIPKIT can leave behind when a member’s residues all fall in trimmed columns, and which FAMSA cannot realign. (by @vagkaratzas, bug reported by @Yixuan39)

Changed

• #186 - Raw family HMMs are now published under a subfolder named after the tool that built them, hmm/raw/hmmer_hmmbuild/ or hmm/raw/mgnifam/, matching how the seed and full MSA outputs are already organised. (by @vagkaratzas, suggested by @nickp60)
• #186 - Restricted samplesheet sample names to letters, digits, dots, underscores and dashes. Family names are parsed back out of file names with split("<sample>_"), which takes a regex, so a sample name carrying a regex metacharacter could mis-split; the sample name also builds output paths. (by @vagkaratzas, raised by @erikrikarddaniel)
• #182 - Updated CHUNK_CLUSTERS and MERGE_SEEDS to optionally output TSV, required by the iterative family generation subworkflow. (by @vagkaratzas)
• #179 - nf-core tools template update to 4.0.3. (by @vagkaratzas)

Dependencies

Changed

• #170 - Updated local modules to versions topic output. (by @vagkaratzas)
• #171
  • Updated nf-core modules and subworkflows to latest, removing all remaining ch_versions. (by @vagkaratzas)
  • Updated pipeline-level nf-schema to 2.7.2. (by @vagkaratzas)

Dependencies

Added

• #159 - Added functionality to generate a HMM library file (compressed) with its respective final protein families HMMs, per each input sample row. Family library files can be found at hmm/library (by @juanfmx2) (Hackathon 2026)
• #154 - Added optional save parameters for update_families mode: --save_update_families_pre_clipped_fasta, and --save_update_families_clipped_fasta (with gaps removed) to save FASTA files from updated family MSAs at various stages of the subworkflow (update_families/fasta/pre_clipped/, update_families/fasta/pre_clipped_non_redundant_sequences/, and update_families/fasta/post_clipped/). (by @eparisis) (Hackathon 2026)

Changed

• #162 - nf-core tools template update to 4.0.2. (by @vagkaratzas)
• #154 - Moved update_families non-redundant sequence FASTA output from mmseqs/update_families/non_redundant_sequences/ to update_families/fasta/pre_clipped_non_redundant_sequences/${meta.id}/ (now controlled by --save_update_families_pre_clipped_fasta). (by @eparisis) (Hackathon 2026)
• #153 - Typo fixes and unused boilerplate removal. (by @vagkaratzas)
• #152
  • Removed an always-truthy if condition. (by @vagkaratzas)
  • Moved a wrongly placed .first() to now properly report all sample representative sequences on the MultiQC report. (by @vagkaratzas)
• #150 - Pipeline logos and workflow maps updated. (by @vagkaratzas)
• #149 - Updated the nf-core/proteinfamilies article citation to the GigaScience publication. (by @vagkaratzas)
• #147 - Modules, subworkflows and pipeline code updates to fix linting warnings and errors. (by @vagkaratzas)

Dependencies

Added

• #143
  • Added the cmaple module for optional phylogenetic tree inference for final family full MSAs. (by @vagkaratzas)
  • Added extra nf-tests for the REMOVE_REDUNDANCY subworkflow. (by @vagkaratzas)
• #140 - Using the new workflow output syntax to publish the downstream nf-core/proteinannotator samplesheet. (by @vagkaratzas)

Changed

• #143 - Updated metro-maps, citations, README.md and output.md to include cmaple phylogenetic trees. (by @vagkaratzas)
• #141 - Aligning the default output directory of the new output syntax with the publishing directory of the pipeline. (by @vagkaratzas)
• #140
  • Updated metro-map to also depict the optional creation of downstream samplesheets for nf-core/proteinfold and nf-core/proteinannotator. Also swapped seqkit/rmdup and seqkit/replace modules to their proper execution sequence. (by @vagkaratzas)
  • Updated the test_full profile time and memory requirements to avoid AWS failures on release. (by @vagkaratzas)

Fixed

• #143
  • Fixed a bug in REMOVE_REDUNDANCY subworkflow, where the combination of these skip flags --skip_sequence_redundancy_removal true, --skip_additional_sequence_recruiting true and --skip_additional_sequence_recruiting false, would execute HHSUITE_REFORMAT_FILTERED to reformat Stockholm alignments while they were already in fasta (or clipkit) format. (by @vagkaratzas)
  • Fixed a bug in REMOVE_REDUNDANCY subworkflow, where unmerged similar families would be removed along with redundant ones, when --skip_family_merging was true and --skip_family_redundancy_removal was false. (by @vagkaratzas)

Dependencies

Added

• #133 - Using the new workflow output syntax to publish the downstream nf-core/proteinfold samplesheet. (by @vagkaratzas)
• #132 - Added optimized memory and time resources for test and test_full profiles. (by @vagkaratzas)

Changed

• #136 - Based on protein family reproducibility benchmarks, cluster is now the default MMseqs2 mode, due to its increased sensitivity compared to linclust. (by @vagkaratzas)
• #135 - nf-core tools template update to 3.5.1. (by @vagkaratzas)

Dependencies

Special Thanks

To @jfy133 and @JoseEspinosa for their assistance with all queries regarding chaining nf-core/proteinfamilies to nf-core/proteinfold.

Added

• #124
  • Added new subworkflow MERGE_FAMILIES that can optionally merge similar (but not redundant) generated protein families. (by @vagkaratzas)
  • Added new functionality to the local module IDENTIFY_REDUNDANT_FAMS which now also detects and outputs the identifiers of similar families that can optionally be merged downstream. These identifiers are written to “/remove_redundancy/<samplename>/similar_fam_ids.txt”, and the corresponding family pairwise similarity scores to “/remove_redundancy/<samplename>/similarities.csv”. (by @vagkaratzas)
  • Added new local module POOL_SIMILAR_COMPONENTS that generates family clusters, from a family-similarity edgelist. (by @vagkaratzas)
  • Added new local module MERGE_SEEDS that merges seed alignments of similar families, before restarting the family generation subworkflow. (by @vagkaratzas)
• #118
  • Added preprint citation to the repo. (by @vagkaratzas)
  • Added separate metro map files for dark and light browser modes. (by @vagkaratzas)
  • Added new local module EXTRACT_FAMILY_MEMBERS which outputs a two-column TSV file containing the final family identifiers and their corresponding member sequence identifiers. The file is saved at “/family_reps/<samplename>/<samplename>.tsv”. (by @vagkaratzas)
• #117
  • Added SEQKIT_SEQ for optional sequence preprocessing in the quality check subworkflow. (by @vagkaratzas)
  • Added SEQKIT_REPLACE for optional sequence name parsing in the quality check subworkflow. (by @vagkaratzas)
  • Added SEQKIT_RMDUP for optional removal of duplicate names and sequences in the quality check subworkflow. (by @vagkaratzas)

Changed

• #128 - nf-core tools template update to 3.4.1.
• #124
  • Conditional workflow flags switched to their skip opposites; --trim_msa to --skip_msa_trimming, --recruit_sequences_with_models to --skip_additional_sequence_recruiting, --remove_family_redundancy to --skip_family_redundancy_removal, --remove_sequence_redundancy to --skip_sequence_redundancy_removal. (by @vagkaratzas)
• #118
  • Swapped the local CHECK_QUALITY subworkflow with the new nf-core one FAA_SEQFU_SEQKIT. (by @vagkaratzas)
  • Based on protein family reproducibility benchmarks (i.e., computationally reproducing manually curated protein family resources), the cluster_seq_identity and cluster_coverage parameter default values have been updated to 0.3 and 0.5 (down from 0.5 and 0.9) respectively. (by @vagkaratzas)
• #117 - Swapped the local SEQKIT_STATS and the local SEQKIT_STATS_TO_MQC modules with the SEQFU_STATS one, which runs a bit faster and produces a MultiQC-ready output without the need for manual parsing. (by @vagkaratzas)

Dependencies

Deprecated

• #124 - Deprecated --trim_msa, --recruit_sequences_with_models, --remove_family_redundancy and --remove_sequence_redundancy. (by @vagkaratzas)

Special Thanks

To @jfy133, @erikrikarddaniel and @chrisAta for this version’s PR code reviews.

Fixed

• #112 - Fixed a bug in EXTRACT_FAMILY_REPS, where all sequences were pasted into the family representative one, and updated the relevant local nf-test. (by @vagkaratzas)

Changed

• #106 - Swapped the local EXECUTE_CLUSTERING subworkflow with the new nf-core MMSEQS_FASTA_CLUSTER one. (by @vagkaratzas)

Dependencies

Changed

• #104 - Pulling params from local subworkflows into main workflow.
• #103 - Parallelized execution for the EXTRACT_FAMILY_REPS local module and changed its input from full_msa to fasta.
• #100 - CAT_CAT module replaced with FIND_CONCATENATE to avoid large scale Argument list too long errors.
• #98 - nf-core tools template update to 3.3.2.

Added

• #105 - CHECK_QUALITY subworkflow added at the start of the pipeline. It utilizes the seqkit/stats nf-core module to generate a MultiQC-ready report with statistics for the input amino acid sequences. The metro-map has been updated to reflect this change.

Added

• #93
  • Added nf-test and meta.yml file for local subworkflow GENERATE_FAMILIES.
  • Added nf-test and meta.yml file for local subworkflow REMOVE_REDUNDANCY.
  • Added nf-test and meta.yml file for local subworkflow UPDATE_FAMILIES.
• #88
  • Added nf-test and meta.yml file for local module BRANCH_HITS_FASTA.
  • Added nf-test and meta.yml file for local module FILTER_NON_REDUNDANT_FAMS.
  • Added nf-test and meta.yml file for local module IDENTIFY_REDUNDANT_FAMS.
  • Added nf-test and meta.yml file for local module EXTRACT_FAMILY_REPS.
  • Added the default pipeline end-to-end nf-test.

Changed

• #81 - nf-core tools template update to 3.3.1.

Fixed

• #80 - Fixed a bug where, due to a missing check for equal family sizes, non-redundant families were erroneously marked as redundant through transitive relationships and were removed

Changed

• #77 - Default branch changed from master to main.
• #73 - Changed the fasta parsing library of the CHUNK_CLUSTERS local module, from pyfastx back to the latest version of biopython, and parallelized its writing mechanism, achieving decreased execution time.

Dependencies

Removed

• #73 - Deprecated pyfastx module version of CHUNK_CLUSTERS, since it was struggling performance-wise with larger datasets.

Added

• #69 - Added the hhsuite/reformat nf-core module to reformat .sto alignments to .fas when in-family sequence redundancy is not removed. Also added the option to save intermediate and final family fasta files throughout the workflow with various save parameters.
• #58 - Added nf-test and meta.yml file for local module REMOVE_REDUNDANCY_SEQS (Hackathon 2025)
• #56 - Added nf-test and meta.yml file for local module FILTER_RECRUITED (Hackathon 2025)
• #55 - Added nf-test and meta.yml file for local module CHUNK_CLUSTERS (Hackathon 2025)
• #54 - Added nf-test for local subworkflow ALIGN_SEQUENCES (Hackathon 2025)
• #53 - Added nf-test for local subworkflow EXECUTE_CLUSTERING (Hackathon 2025)
• #51 - Added nf-test and meta.yml file for local module CALCULATE_CLUSTER_DISTRIBUTION (Hackathon 2025)
• #34 - Added the EXTRACT_UNIQUE_CLUSTER_REPS module, that calculates initial MMseqs clustering metadata, for each sample, to print with MultiQC (Id,Cluster Size,Number of Clusters)

Fixed

• #69 - Fixed a bug where redundant family alignments were not published properly, if intra-family redundancy removal mechanism was switched off #68
• #65 - Fixed a bug in CHUNK_CLUSTERS, where pipeline would crash if the module filtered out all clusters, due to a high membership threshold #64
• #35 - Fixed a bug in remove_redundant_fams.py, where comparison was between strings instead of integers to keep larger family
• #33 - Fixed an always-true condition at the filter_non_redundant_hmms.py script, by adding missing parentheses
• #29 - Fixed hmmalign empty input crash error, by preventing the FILTER_RECRUITED module from creating an empty output .fasta.gz file, when there are no remaining sequences after filtering the hmmsearch results #28

Changed

• #69 - Changed the publish directory architecture for HMMs, seed MSAs, full MSAs and family FASTA files, to make it more intuitive. REMOVE_REDUNDANT_FAMS local module converted to IDENTIFY_REDUNDANT_FAMS to extract redundant family ids which will then be used downstream. FILTER_NON_REDUNDANT_HMMS local module converted to FILTER_NON_REDUNDANT_FAMS and reused four times (HMM, seed MSA, full MSA, FASTA). Changed the output format of the EXTRACT_FAMILY_REPS and REMOVE_REDUNDANT_SEQS local modules from .fa to .faa. Metro map updated with new hhsuite/reformat module.
• #57 - slight improvements of nextflow_schema.json (Hackathon 2025)
• #57 - slight improtmenets of assets/schema_input.json (Hackathon 2025)
• #34 - Swapped the SeqIO python library with pyfastx for the CHUNK_CLUSTERS module, quartering its duration
• #32 - Updated ClipKIT 2.4.0 -> 2.4.1, that now also allows ends-only trimming, to completely replace the custom CLIP_ENDS module. Users can now also define its output format by setting the --clipkit_out_format parameter (default: clipkit)

Dependencies

Deprecated

• #32 - Deprecated CLIP_ENDS module and --clipping_tool parameter. The only option now is ClipKIT, covering both previous modes, via setting --trim_ends_only

Initial release of nf-core/proteinfamilies, created with the nf-core template.

Added

• Amino acid sequence clustering (mmseqs)
• Multiple sequence alignment (famsa, mafft, clipkit)
• Hidden Markov Model generation (hmmer)
• Between families redundancy removal (hmmer)
• In-family sequence redundancy removal (mmseqs)
• Family updating (hmmer, seqkit, mmseqs, famsa, mafft, clipkit)
• Family statistics presentation (multiqc)

By @vagkaratzas and @mberacochea.