nf_funannotate1

A Nextflow DSL2 framework for eukaryotic genome annotation (fungal defaults, generic-capable) that runs on a workstation, any SLURM cluster, Kubernetes, or the UCR HPCC. It runs the full funannotate workflow — genome clean → repeat mask → RNA-seq fetch/train → predict → optional antiSMASH/InterProScan/SignalP → annotate/update — plus a standalone EarlGrey curated repeat-masking pipeline. The pipeline runs directly from GitHub; no clone is required.

Install

• Java 17+
• Nextflow 25.10.0+ (curl -s https://get.nextflow.io | bash)
• Apptainer or Singularity on PATH (on every compute node, if you use a cluster)

Quick start

Run from a launch directory that holds samples.csv. Outputs, caches and lib/ assets live in the launch directory, not in Nextflow's cached copy of the repo (~/.nextflow/assets/). Add -r <branch|tag> to pin a revision.

# Check the workflow graph first (synthetic data, no real tools):
nextflow run stajichlab/nf_funannotate1 -profile test -stub-run

# SLURM cluster (mysite.config holds your partitions and paths):
nextflow run stajichlab/nf_funannotate1 -profile annotate,slurm,singularity -c mysite.config -resume

# Workstation or single server:
nextflow run stajichlab/nf_funannotate1 -profile annotate,local,singularity -resume

On a cluster, run the Nextflow head process in a small, long job (about 2 CPUs and 32 GB of RAM, because image pulls run in the head process):

sbatch -c 2 --mem 32G --time 7-00:00:00 --wrap \
  "nextflow run stajichlab/nf_funannotate1 -profile annotate,slurm,singularity -c mysite.config -resume"

Minimal samples.csv (one row per genome; GENOME is a FASTA path, plain or gzipped):

SPECIES,STRAIN,ASMID,LOCUSTAG,BUSCO_LINEAGE,TRANSL_TABLE,NCBI_TAXONID,GENOME
Aspergillus nidulans,FGSC-A4,ANID_A4,ANIDA4,eurotiomycetes_odb10,1,227321,genomes/Aspergillus_nidulans_FGSC-A4.fasta

Minimal mylab_params.yaml (pass with -params-file; use absolute paths):

sif_dir: /shared/containers          # container images and Apptainer pull cache
# proteins: /data/db/uniprot_sprot_fungi.faa   # default: <launch dir>/lib/swissprot_fungi.faa

RNA-seq comes from one of two sources, chosen per species (normally not mixed):

• Your own reads: place <Species_name>_norm_R1/R2/SE.fastq.gz and a header-only sra_query/<Species_name>.sra_query.csv plus its .key in rnaseq_reads/. Nothing is downloaded.
• Public reads from NCBI SRA: leave rnaseq_reads/ empty for that species. The pipeline queries SRA, downloads and normalizes the reads into rnaseq_reads/.

Exact file names and an example are in README_Extended.md. To test one genome first add --n_test 1; to pick one use --asmid ANID_A4. After a stop or failure, run the same command again and -resume continues. The full walkthrough (read layout, samples.csv columns, params, container pulls) is in README_Extended.md.

Profiles

Compose one option from each axis: -profile <pipeline>,<executor>,<provisioning>.

Axis	Options
pipeline	annotate · earlgrey · test / stub
executor	slurm · local
provisioning	singularity (containers; portable) · pixi · conda (shared frozen envs) · ucr_hpcc (UCR Lmod modules; institutional example)

• Your own SLURM cluster: put partitions, account and the container cache in a small mysite.config (start from conf/site_template.config) and pass it with -c. See docs/adding_a_site.md and Running on your own SLURM cluster.
• Kubernetes (NRP Nautilus): -profile annotate,nrp replaces the executor and provisioning picks. Any NRP user can use it, not only UCR. See k8s/README.md.
• Two clusters at once (for example SLURM plus NRP): give each its own samples.csv slice and launch directory. See docs/multi_cluster.md.

UCR HPCC

UCR HPCC is the institutional example, not a requirement. Its Lmod profile, launcher auto-detection and shared paths are in docs/site_ucr_hpcc.md.

Outputs

In the launch directory: input_clean_genomes/ (cleaned and masked genomes), rnaseq_data/ (Trinity transcripts), genome_annotation_training/, and genome_annotation/<Species_Strain>/predict_results/ (GFF3, GBK, protein and transcript FASTA), plus funannotate_db/ and work/ caches (keep them for -resume).

More documentation

• README_Extended.md — full walkthrough and reference:
  • Quick start: local genomes + RNA-seq
  • Orthogonal profiles and funannotate version / EVM selection
  • Singularity images and PASA MySQL backend
  • Throughput, resumability & storage
  • EarlGrey repeat masking
  • Layout, Testing
• docs/usage.md — usage and parameters
• docs/output.md — full list of outputs
• docs/adding_a_site.md — SLURM site configuration
• docs/multi_cluster.md — splitting one batch across clusters (e.g. SLURM + NRP)
• docs/site_ucr_hpcc.md — UCR HPCC as a worked example site
• CHANGELOG.md · CITATIONS.md