- CGE / isolate typing extensions:
disinfinder,pmlst,mlst,abricate - References 136–138
- Expanded
cge-findersrouting for biocide genes, plasmid/chromosomal ST, and multi-DB mass screens - Skills catalog total: 163; package version 1.21.0
- CGE Finder layer:
cge-findershub,plasmidfinder,resfinder,pointfinder,virulencefinder,mobileelementfinder,staramr - References 129–135
- Skills catalog total: 159; package version 1.20.0
- Wired CGE Finders into microbial-mining / tool-selection / docs
- Polished routing docs: README, getting-started, docs/tool-selection aligned with research-analysis / LM / mining hubs
- Clarified OMG →
genome-language-model; Foldseek family → homology-search - Catalogued
buscoin docs/skills.md; refreshed workflow/tool-selection blurbs - Package version 1.19.1 (152 skills)
- Research analysis extensions:
ancombc,sparcc,spieceasi,stamp,phyloseq,nextflow,snakemake - References 122–128
- Skills catalog total: 152; package version 1.19.0
- Research analysis layer:
research-analysishub,microeco,maaslin2,lefse,multiqc; deepenedvisualization - References 118–121
- Skills catalog total: 145; package version 1.18.0
- DOM / biosurfactant layer:
dom-biosurfactanthub,hadeg,biosurfdb - Deepened
dbcanfor glycan DOM substrate / CGC prediction - References 115–117
- Skills catalog total: 140; package version 1.17.0
- Element cycling layer:
element-cyclinghub,metabolic(METABOLIC),mebs,fegenie(C/N/S/Fe/P/H₂/CH₄/…) - Defenseome:
padloc(with DefenseFinder) - MGE refinement:
mobileog,isescan,integronfinder,mob-suite - Quorum sensing:
qsap - References 107–114
- Expanded
microbial-mining/tool-selectionrouting - Skills catalog total: 137; package version 1.16.0
esmc(ESM C / Cambrian) andsaprot(Westlake SaProt) protein LM skills- References 105–106
- Skills catalog total: 127; package version 1.15.0
- Protein language model layer:
protein-language-modelhub,esm(ESM-2),prostt5(ProstT5 / ProtT5) - References 102–104
- Routed
plmsearch/deepblast/ LucaPhylo viaprotein-language-model - Skills catalog total: 125; package version 1.14.0
dnabert2,caduceusgenome language model skills- References 100–101
- Renamed hub
genomic-lm→genome-language-model(display: Genome language model) - Skills catalog total: 122; package version 1.13.0
- Genome language model layer:
genome-language-modelhub,nucleotide-transformer(NT/NTv3),evo2 - Deepened
omg/gLM2 notes; references 96–99
- Skills catalog total: 120; package version 1.12.0
colabfoldskill (deferred; use AFDB / Foldseek+ProstT5 paths instead)
- Skills catalog total: 117; package version 1.11.1
- Extended Steinegger/Söding search stack:
hh-suite,colabfold,plass,metaeuk,foldcomp,foldmason,petasearch,unicore - Expanded
homology-search/phylogenomicsrouting
- Skills catalog total: 118; package version 1.11.0
- Homology-search layer:
homology-searchhub,erast(vector DB),phylign(phylogenetic compression / MiniPhy+Phylign) - Phylogenomics layer:
phylogenomicshub,iqtree,fasttree,phylophlan,lucaphylo - References 84–89
- Wired search/tree routing into tool-selection, workflow, mining, LLM hub, docs
- Skills catalog total: 110; package version 1.10.0
- Steinegger structure-search layer:
foldseek(3Di search, Multimer, ProstT5, clustering),folddisco(structural motifs) - References 80–83 (Foldseek, Foldseek-Multimer, Folddisco, AFDB clustering)
- Expanded
metagenomics-llmrouting (sequence → fold → multimer → motif) - Deepened
mmseqs2as sequence counterpart; cross-links in mining/tool-selection - Skills catalog total: 102; package version 1.9.0
- AI / language-model layer:
metagenomics-llm,omg(OMG+gLM2),plmsearch,deepblast(TM-Vec),spacedust,alphagem - Docs/routing hooks in public-databases, tool-selection, microbial-mining, README
- References 74–79
- Skills catalog total: 100; package version 1.8.0
- Quality/consistency pass: synced pipeline, workflow, tool-selection, README, getting-started, and troubleshooting with the full skill set
- Removed misleading ensemble-binning Decision notes from non-binning skills
- Deepened high-traffic skills: kaiju, maxbin2, prokka, bakta, eggnog-mapper, rgi, mmseqs2, quast, salmon, mag-qc
- Cross-linked SingleM/Bin Chicken, vConTACT/CheckV, Deepurify/Anvi'o, ANI tools
- Duplicate Related skills headers introduced during cleanup
- Priority skills:
anvio,singlem,vcontact,deepurify,fastani - Secondary skills:
vmh,insilicoseq,plasmaag,aamb,taxvamb,metabinner,lexicmap - skani↔FastANI comparison; mining/tool-selection/public-data routes updated
- References 61–73
- Skills catalog total: 94; package version 1.6.0
- Data-layer skills:
cami(CAMI/CAMI II + CAMISIM/AMBER/OPAL),progenomes,kegg - Expanded
public-databaseshub anddocs/public-databases.md(four data layers) - GTDB release 10 notes; richer MGnify catalogue guidance
- References 51–60 (CAMI stack, GTDB R10, proGenomes4, KEGG, profiler abundance types)
- Skills catalog total: 82; package version 1.5.0
- New skills from curated literature:
lorbin,comebin,binchicken,metabuli - Literature-grounded expansions for ensemble binning, QC, taxonomy, coverage, strain, and viral QC skills (BASALT, MetaWRAP, DAS Tool, SemiBin2, VAMB, CoverM, CheckM2, GUNC, MetaPhlAn, Kraken2/Bracken, mOTUs, sylph, inStrain, CheckV, CONCOCT, GTDB-Tk, DIAMOND)
- Study-design notes on multi-coverage binning and targeted coassembly
- References 39–50 in
docs/references.md
- Tool-selection routes updated for long-read binning, Metabuli, Bin Chicken
- Skills catalog total: 79
- Deepened mapping skills:
bowtie2,strobealign,bwa-mem2,samtools - Deepened mining/binning:
antismash,genomad,dram,metabat2 - New skills:
racon,pilon,skani,dbcan,bigscape - Expanded troubleshooting and compute-setup by stage
- Tool-selection / pipeline / references updated for polish, ANI, CAZyme, BGC networks
- Mapping-stage tools (Bowtie2, BWA, BWA-MEM2, Minimap2, strobealign, Samtools, …)
- Single binners (MetaBAT2, MaxBin2, CONCOCT, SemiBin2, VAMB) and long-read assemblers
- Microbial mining skills:
microbial-mining,antismash,genomad,virsorter2,checkv,macrel,dram,bakta,deeparg,gapseq,defensefinder - Nature-style
docs/references.mdentries for mapping and mining papers
- Pipeline framed as QC → assembly → mapping → binning → taxonomy → function → mining
- Removed third-party list branding from skill/docs surfaces
- Public database skills:
public-databases,sra-ena,mgnify,gtdb,biobakery-databases - Ensemble binning peer skill:
dastool - Docs:
docs/public-databases.md
- Reframed the package around five stages: QC → assembly → binning → taxonomic annotation → functional annotation (bins/MAGs are binning outcomes, not the sole pipeline axis)
- Positioned
basalt,metawrap, anddastoolas an ensemble binning family with BASALT recommended - Updated catalogs, workflow, README, SVG, and manifests (30 skills)
- Per-software analysis skills with upstream GitHub links and analytical-thinking sections:
fastp,kneaddata,metaphlan,humann,kraken2,megahit,metaspades,prodigal,cd-hit,salmon,eggnog-mapper,rgi,metawrap,drep,coverm,gtdbtk,checkm2(plus existingbasalt) - Expanded
basaltskill against the official BASALT guide CLI
- Replaced composite stage skills (
preprocessing,read-based,gene-catalogue,mag-classical) with tool-level skills - Updated catalogs, pipeline docs, and manifests for 24 skills total
- Initial Metagenomics Skills package (Agent Skills layout)
- Multi-platform manifests and
scripts/install-platforms.sh