merged_annotator

A Nextflow pipeline for genome and metagenome annotation. Combines structural annotation (Bakta / Pyrodigal / Prodigal / Prokka), functional annotation (EggNOG-mapper, KofamScan, dbCAN, VFDB), and optional ARG screening (AMRFinderPlus, fARGene, RGI, DeepARG, ABRicate).

---

Quick start

nextflow run main.nf -profile standard --mode single --manifest samples.csv

The manifest is a CSV with at minimum two columns:

ID,assembly
sample1,/path/to/sample1.fasta
sample2,/path/to/sample2.fasta

---

Modes

The --mode parameter controls which annotation tool is used and sets sensible defaults for the rest of the pipeline. There are two named presets and four direct tool selectors.

Named presets

--mode single (default)

For finished or high-quality draft isolate genomes.

• Annotator: Pyrodigal (-p single, fast ORF prediction)
• Protein clustering: off

nextflow run main.nf -profile standard --mode single --manifest samples.csv

--mode meta

For metagenome-assembled genomes (MAGs) or any set of assemblies to be treated as a comparative/metagenomic dataset.

• Annotator: Pyrodigal (-p meta, fast ORF prediction)
• Protein clustering: on (MMseqs2 easy-linclust before functional annotation)

nextflow run main.nf -profile standard --mode meta --manifest samples.csv

Direct tool selectors

Pass the tool name as --mode to bypass the presets entirely. Pipeline-level switches (cluster_proteome) remain at their defaults (off) unless you also set them.

--mode	Annotator	-p flag
bakta	Bakta	n/a
pyrodigal	Pyrodigal	-p single
prodigal	Prodigal	-p single
prokka	Prokka	n/a

# Prokka with ARG screening
nextflow run main.nf -profile standard --mode prokka --manifest samples.csv --arg_annotate

# Pyrodigal with protein clustering
nextflow run main.nf -profile standard --mode pyrodigal --manifest samples.csv --cluster_proteome

---

Key parameters

Parameter	Default	Description
--manifest	—	Path to input CSV (required)
--outdir	results	Output directory
--mode	single	Run mode (see above)
--arg_annotate	false	Run ARG screening subworkflow
--func_annotate	false	Run Functional screening subworkflow
--cluster_proteome	false	Cluster proteins with MMseqs2 before functional annotation

Database paths (pre-configured at Sanger)

Parameter	Default path
--annotation_bakta_db	/data/pam/software/bakta/v6.0/
--arg_abricate_db	/data/pam/software/abricate/db/
--arg_amrfinderplus_db	/data/pam/software/amrfinder/latest/
--eggnog_data_dir	/data/pam/software/eggnog/v5.0/
--dbcan_db	/data/pam/software/run_dbcan/5.2.5/db/
--vfdb_db	/data/pam/team230/sd28/scratch/secondment_162/dbs/vfdb/VFDB_setB_pro.dmnd
--kofam_db	/data/pam/software/kofam_scan/

Key thresholds

Parameter	Default	Description
--module_completeness	0.5	KEGG module completeness threshold
--kofamscan_eval	0.00001	KofamScan e-value cutoff
--vfdb_identity	50	VFDB minimum identity (%)
--vfdb_coverage	80	VFDB minimum subject coverage (%)
--vfdb_e_value	1e-10	VFDB e-value cutoff
--cazyme_hmm_eval	default	dbCAN HMM e-value cutoff
--cazyme_hmm_cov	default	dbCAN HMM minimum coverage
--mmseqs_min_id	0.8	MMseqs2 minimum sequence identity
--mmseqs_min_cov	0.9	MMseqs2 minimum coverage

---

Advanced usage

Tool parameters (translation table, HMM thresholds, per-tool ARG flags etc.) are defined with sensible defaults in conf/advanced.config. Override them on the CLI as needed — no need to edit the file directly.

To see a full list of these params use the --helpFull command

nextflow run main.nf --helpFull

Running annotation tools in meta mode via direct tool selector

When using a direct tool selector (--mode pyrodigal, --mode prodigal, --mode bakta) the tools default to single-genome mode. Use the corresponding flag to force meta mode:

--mode	Override flag	Effect
pyrodigal	--annotation_pyrodigal_meta	-p meta
prodigal	--annotation_prodigal_meta	-p meta
bakta	--annotation_bakta_meta	--meta

nextflow run main.nf -profile standard --mode pyrodigal --manifest samples.csv \
    --annotation_pyrodigal_meta

nextflow run main.nf -profile standard --mode bakta --manifest samples.csv \
    --annotation_bakta_meta

Enabling/disabling individual ARG tools

When --arg_annotate, all five tools run by default. Skip individual ones:

nextflow run main.nf -profile standard --mode single --manifest samples.csv \
    --arg_annotate \
    --arg_skip_deeparg \
    --arg_skip_fargene

---

Output structure

results/
  annotation/
    pyrodigal/<sample_id>/       # or bakta/, prodigal/, prokka/
  func/
    kofamscan/
    dbcan/
    vfdb/
    eggnogmapper/
  arg/                       # only when --arg_annotate
    amrfinderplus/<sample_id>/
    abricate/<sample_id>/
    rgi/<sample_id>/
    deeparg/<sample_id>/
    fargene/<sample_id>/
    hamronization/
    argnorm/
  mmseqs/                    # only when --cluster_proteome
  reports/
    hamronization_summarize/