Sign inSign up

albertosola/novum_pipeline

By albertosola

•Updated 16 days ago

A novel bioinformatics strategy to predict and annotate small RNAs-target interactions.

Image
Data science
0

98

albertosola/novum_pipeline repository overview

⁠Novum Pipeline

Novum Pipeline predicts which bacterial genes a set of miRNAs may bind. It runs two RNA–RNA interaction tools, RNAhybrid and IntaRNA, over the same miRNAs and genomes, and reports each tool's predictions plus the consensus: the miRNA–gene pairs both tools predict.

This image holds the whole Snakemake⁠ workflow, with every tool installed and its version pinned. It runs the same on Linux, WSL and Macs, and needs no conda setup. From your FASTA files, one command produces:

  • annotated tables, and readable reports with each duplex drawn out;
  • RNAhybrid p-values calibrated for each miRNA and gene length, identical on every rerun;
  • IntaRNA predictions with or without site accessibility;
  • a length-corrected energy for ranking, because longer genes score better by chance;
  • optional PDF plots.

Source code and full documentation: github.com/GenRegDSCB/novum_pipeline⁠

⁠Quick start

Pull the image, a 1.4 GB download that takes 5.7 GB on disk:

docker pull --platform linux/amd64 albertosola/novum_pipeline:v1.0.0

Save the run command as an alias. Add the line to ~/.bashrc (~/.zshrc on a Mac) to keep it:

alias novum='docker run --rm -it --platform linux/amd64 -u "$(id -u):$(id -g)" -v "$PWD":/work albertosola/novum_pipeline:v1.0.0'

In an empty project folder, write a starter config:

mkdir my_project && cd my_project
novum init

In Config/config.yaml, replace my_sample under queries: (your miRNA FASTA) and targets: (your CDS FASTA) with your own files. Keep the files inside the project folder, for example under Data/Raw/: the container sees nothing else. Then run:

novum --cores 8

Results go to Data/Results/<sample>/. Read *_enhanced.txt to go through hits one by one, and use *_annotated.csv to sort or filter them. Outputs⁠ describes every file and column, and Configuration⁠ every setting.

⁠Usage

Whatever follows novum, or the image name in a full docker run, decides what runs:

After the image nameWhat it does
initWrites a starter Config/config.yaml, with results in Data/Results/. It never overwrites an existing config.
prepare-inputs --pairs pairs.txtDownloads the genomes from NCBI and the miRNAs from miRBase for a list of (taxon, miRNA) pairs, and replaces the samples in the config. Run init first. See Preparing inputs⁠.
anything elsePassed to Snakemake: -n for a dry run, --cores 8, or file names to run one sample. See Running⁠.

What the alias's options do:

  • -v "$PWD":/work shares the current folder, the only one the container sees.
  • -u "$(id -u):$(id -g)" makes you, not root, the owner of the files it writes.
  • -it shows progress and lets Ctrl+C stop the run. --rm deletes the container when it stops; your files stay.

Long runs. Run detached, with -d --name novum in place of the alias's --rm -it:

docker run -d --name novum --platform linux/amd64 -u "$(id -u):$(id -g)" -v "$PWD":/work albertosola/novum_pipeline:v1.0.0 --cores 8
docker logs -f novum     # Ctrl+C stops following, not the run
docker rm novum          # after it finishes, before the next detached run

docker stop novum stops the run. To resume an interrupted run, run the same command again: Snakemake keeps every finished step. If it reports incomplete files, add --rerun-incomplete.

Older results. Snakemake reruns from the start any results folder made before v1.0.0, or by the conda install, calibration included. Results made before v1.0.0⁠ shows how to keep such a folder as it is.

NCBI API key. A key raises NCBI's limit for prepare-inputs from about 3 to 10 requests per second. Set it as NCBI_API_KEY. The alias does not pass it to the container, so add -e NCBI_API_KEY before the image name in the full command.

Apple Silicon Macs. The image is built for Intel/AMD only, because RNAhybrid has no ARM build. Docker Desktop runs it through Rosetta automatically, more slowly.

CPUs and memory.

  • Give --cores a number. With all, Snakemake ignores a Docker --cpus limit and counts every core.
  • Large panels need memory: the RNAhybrid arm has needed about 12 GB. Docker Desktop on a Mac gives containers half the Mac's memory by default; raise it under Settings → Resources. WSL gives Linux half of Windows' memory; raise it in %UserProfile%\.wslconfig.

Clusters without Docker usually have Apptainer, which runs the same image. Pull it once on a node with internet access, then run it from the project folder:

apptainer pull novum.sif docker://albertosola/novum_pipeline:v1.0.0
apptainer run --cleanenv --bind "$PWD":/work --pwd /work novum.sif --cores 8

--cleanenv keeps your shell's variables out of the pipeline's environments.

⁠Tags

Image vX.Y.Z is built from git tag vX.Y.Z, and latest is the newest. Name a version for reproducible work: latest changes with each release.

⁠License and citation

MIT. If you use this workflow in a publication, please cite the repository and the tools it relies on; Citation⁠ lists them.

Tag summary

Content type

Image

Digest

sha256:b442fb8c0…

Size

1.3 GB

Last updated

16 days ago

docker pull albertosola/novum_pipeline