Seblastian

Seblastian predicts eukaryotic selenoprotein genes by searching genomic sequences for SECIS elements and then analysing the upstream regions with BLAST and Exonerate.

The workflow was described in:

Mariotti M, Lobanov AV, Guigó R, Gladyshev VN. SECISearch3 and Seblastian: new tools for prediction of SECIS elements and selenoproteins. Nucleic Acids Research. 2013;41(15):e149. https://doi.org/10.1093/nar/gkt550

Seblastian is legacy software written for Python 2.7. Linux x86-64 is the best-supported platform, particularly for the bundled SECISearch and COVE executables.

SECISearch and sEBLASTIAN are also available through our web server.

Installation with Conda

Install Conda or Mamba, then clone this repository:

git clone https://github.com/mariottigenomicslab/Seblastian.git
cd Seblastian

Create and activate an environment containing Python 2.7 and the external command-line dependencies:

conda create -n seblastian \
  --channel conda-forge \
  --channel bioconda \
  --strict-channel-priority \
  python=2.7 \
  perl \
  gawk \
  blast-legacy=2.2.26 \
  exonerate=2.4.0 \
  infernal=1.0.2 \
  viennarna=2.4.18

conda activate seblastian

ViennaRNA 2.4.18 is used here because it is the latest Bioconda build compatible with Python 2.7. Make the bundled scripts discoverable and create a writable temporary directory:

export SEBLASTIAN_HOME="$(pwd)"
export PATH="$SEBLASTIAN_HOME:$SEBLASTIAN_HOME/bin:$PATH"
chmod +x "$SEBLASTIAN_HOME/blaster_parser.g"
mkdir -p "$SEBLASTIAN_HOME/tmp"

Display the command-line help:

python "$SEBLASTIAN_HOME/Seblastian.py" -h

The environment variable and PATH change apply to the current shell. Set them again after opening a new terminal.

Running Seblastian

SECIS search only

mkdir -p results

python "$SEBLASTIAN_HOME/Seblastian.py" \
  -t target.fa \
  -o results/seblastian \
  -SS \
  -temp "$SEBLASTIAN_HOME/tmp" \
  -infernal_cm "$SEBLASTIAN_HOME/SECIS_infernal.cm" \
  -infernal_stk "$SEBLASTIAN_HOME/SECIS_infernal.stk" \
  -covels_cm "$SEBLASTIAN_HOME/SECIS_covels.cm" \
  -bin_folder "$SEBLASTIAN_HOME/bin"

Full selenoprotein prediction

mkdir -p results

python "$SEBLASTIAN_HOME/Seblastian.py" \
  -t target.fa \
  -o results/seblastian \
  -d "$SEBLASTIAN_HOME/uniref50.only_selenoproteins.fa" \
  -temp "$SEBLASTIAN_HOME/tmp" \
  -infernal_cm "$SEBLASTIAN_HOME/SECIS_infernal.cm" \
  -infernal_stk "$SEBLASTIAN_HOME/SECIS_infernal.stk" \
  -covels_cm "$SEBLASTIAN_HOME/SECIS_covels.cm" \
  -bin_folder "$SEBLASTIAN_HOME/bin"

Replace target.fa with a nucleotide FASTA file. Run python "$SEBLASTIAN_HOME/Seblastian.py" -h to see all available routines and options.

Alternative: Docker

A prebuilt image is available from Docker Hub:

docker pull maxtico/seblastian:latest

Display the help:

docker run --rm maxtico/seblastian:latest \
  python /Seblastian/Seblastian.py -h

Run SECIS prediction on target.fa in the current directory and write the results back to that directory:

docker run --rm \
  --mount type=bind,source="$(pwd)",target=/data \
  maxtico/seblastian:latest \
  python /Seblastian/Seblastian.py \
  -t /data/target.fa \
  -o /data/seblastian \
  -SS

Run the full pipeline with the protein database included in the image:

docker run --rm \
  --mount type=bind,source="$(pwd)",target=/data \
  maxtico/seblastian:latest \
  python /Seblastian/Seblastian.py \
  -t /data/target.fa \
  -o /data/seblastian

Models and protein database

The repository currently includes the resources used by the program:

• SECIS_infernal.cm, SECIS_infernal.1.0.2.cm, and SECIS_covels.cm are covariance models.
• SECIS_infernal.stk is the alignment associated with the Infernal model.
• uniref50.only_selenoproteins.fa is the bundled protein database.
• The .phr, .pin, and .psq files are the prebuilt legacy BLAST indexes for that protein database.

The bundled protein database and its indexes occupy less than 1 MB, so they are kept in this repository. A different protein FASTA database can be supplied with -d; Seblastian will create legacy BLAST indexes when they are absent.

If a future database is too large for GitHub, publish a versioned archive in a data repository such as Zenodo and document its DOI, checksum, and expected filename here.

Contributing

Changes can be proposed through GitHub issues and pull requests. When changing the prediction workflow, include the command and input used to verify the result.

License

See LICENSE.