rnahub

rnahub orchestrates iterative nhmmer/Infernal/R-Scape searches for discovering new RNA homologs.

The CLI wires common structural-RNA tools, organizes job folders, and captures diagnostics so you can re-run specific pipeline stages with minimal bookkeeping.

For the webserver, go to http://rnahub.org

(databases for now are not included)

Cite

Magnus, M., Gao, W., Dutta, N., Vicens, Q., Rivas, E.
RNAhub - An automated pipeline to search and align RNA homologs with secondary structure assessment
Nucleic Acids Research 2025
https://academic.oup.com/nar/article/53/W1/W496/8121641

Highlights

• Launch nhmmer searches (v0–v3) with automatic job-folder management.
• Run optional steps such as R-Scape, Infernal, RepeatMasker, or flanked-search generation without retooling commands.
• Resume work by skipping completed stages, or run in --dry mode to inspect commands before execution.
• Designed for workstation runs as well as scheduled jobs on SLURM clusters.

Installation

Python dependency

Install Python requirements that are not bundled with the standard library. At the moment only icecream is needed:

python -m pip install icecream  # installs into the Python you are using

Configure tool paths

Create config_local.py in the project root to override paths from config.py. Required entries point to your nhmmer, Infernal, R-Scape, and auxiliary binaries:

nhmmer = "/home/rnahub/opt/hmmer/hmmer-3.3.2/bin/nhmmer"
RSCAPE_PATH = "/home/rnahub/opt/rscape/rscape_v2.5.2/bin/R-scape"
EASEL_PATH = "/home/rnahub/opt/hmmer/hmmer-3.3.2/bin/"
INFERNAL_PATH = "/usr/bin/"  # used for cmalign
REPEAT_MASKER_PATH = "/home/rnahub/opt/RepeatMasker"
# Optional Rfam resources
RFAM_FILES = "/home/rnahub/rnahub/rfam"
RFAM_DB_PATH = "/home/rnahub/db/rfam/Rfam.cm"

Any variable omitted from config_local.py falls back to the defaults provided in config.py.

Harvard RC cluster snippet

Example session for the FAS RC environment:

module load python/3.10.12-fasrc01
python -m pip install --user icecream

Usage

1. Prepare an input .fa/.fasta sequence or .sto alignment.
2. (Optional) Pre-filter the search space using BLAST against NR to focus on a taxonomic group. Add flanking regions if needed.
3. Run the pipeline, e.g.:

python rnahub.py \
  --input example/seq.fa \
  --db db/fungi.fasta \
  --job-name seq_run \
  --job-folder jobs/seq_run \
  --cpus 8

Set --slurm to emit SLURM submission scripts, --dry to preview commands, and pass --dev-skip-* switches to resume from intermediate stages (for instance --dev-skip-nhmmer123 to skip later nhmmer rounds).

When re-running inside a finished job folder:

python rnahub.py --job-folder . --dev-skip-nhmmer123

Flanked mode

Add flanks via either:

• --flanked path/to/flanks.fa (FASTA only, not .fasta).
• --flanks-in-header where headers follow ><seq_name> <start>-<end>.

Command-line reference

$ python rnahub.py -h
usage: rnahub.py [-h] [--db DB [DB ...]] [--job-name JOB_NAME] [--verbose] [-v] [--slurm]
                 [--evalue EVALUE] [--evalue-for-flanking-search EVALUE_FOR_FLANKING_SEARCH]
                 [--lmin LMIN] [--evalue-final EVALUE_FINAL]
                 [--iteractions ITERACTIONS] [--cpus CPUS] [--dry] [--repeatmasker] [--utot]
                 [--job-folder JOB_FOLDER] [--dev-skip-search] [--dev-skip-nhmmer0]
                 [--dev-skip-nhmmer123] [--dev-skip-rscape] [--dev-skip-infernal]
                 [--rscape-path RSCAPE_PATH] [--rscape] [--input INPUT [INPUT ...]]
                 [--flanked FLANKED] [--flanks-in-header] [--flanks-start FLANKS_START]
                 [--flanks-end FLANKS_END]

optional arguments:
  -h, --help            show this help message and exit
  --db DB [DB ...]      databases (.cm/.sto) searched by nhmmer
  --job-name JOB_NAME   defaults to the input file stem
  --verbose             increase logging
  -v, --version         print the version based on the latest Git tag
  --slurm               submit to SLURM instead of running locally
  --evalue EVALUE       e-value threshold for all runs except the final one
  --evalue-for-flanking-search EVALUE_FOR_FLANKING_SEARCH
                        e-value threshold for flanking (v0) search; defaults to --evalue
  --lmin LMIN           `esl-alimanip` minimum length (default 50)
  --evalue-final EVALUE_FINAL
                        e-value threshold for the last round
  --iteractions ITERACTIONS
                        number of nhmmer iterations
  --cpus CPUS           nhmmer CPU count
  --dry                 show commands without executing them
  --repeatmasker        run RepeatMasker preprocessing
  --utot                treat sequences as DNA (U→T) for RepeatMasker
  --job-folder JOB_FOLDER
                        destination folder (default `jobs/<name>/<input>`)
  --dev-skip-search     skip v0–v3 nhmmer searches
  --dev-skip-nhmmer0    skip only the v0 nhmmer search
  --dev-skip-nhmmer123  skip v1–v3 nhmmer searches
  --dev-skip-rscape     skip the R-Scape stage
  --dev-skip-infernal   skip Infernal stage
  --rscape-path RSCAPE_PATH
                        override R-Scape path
  --rscape              run only the R-Scape step
  --input INPUT [INPUT ...]
                        FASTA/FA or Stockholm input sequences/alignments
  --flanked FLANKED     supply a FASTA with flanking regions
  --flanks-in-header    parse `><seq start-end` spans from headers
  --flanks-start FLANKS_START
                        left flank coordinate
  --flanks-end FLANKS_END
                        right flank coordinate

Workflow tips

• Run BLAST on NR before nhmmer to scope the search space (e.g., fungi).
• Add flanks to recover context for hits.
• Keep a cache of pre-compiled genomes for faster iterative runs.
• Inspect .helixcov files produced by newer R-Scape versions.

Gao, W., Jones, T. A. & Rivas, E. Discovery of 17 conserved structural RNAs in fungi. Nucleic Acids Res 49, gkab355 (2021). https://academic.oup.com/nar/article/49/11/6128/6292099