nanjiang/pathopred repository overview

⁠docker_pathopred

Docker container for amino acid substitution pathogenicity prediciton.

⁠Installation and usage with Docker

1. Get Docker image

docker pull nanjiang/pathopred

1. Prepare BLAST database. The container uses the NCBI nr database. If you do not have the NCBI nr database available, BLAST distributions provide the update_blastdb.pl Perl script which can be used to obtain the database. You can also download only the script and use it on its own. Move to the folder where you want to place the database, and run

perl update_blastdb.pl --decompress nr

1. Start the docker container in the background. Assuming you have the NCBI nr database at /data/blastdb, start the container with your local user $USER and the localhost directory /scratch mounted to /scratch inside the container by running

docker run -e USER_ID=$(id -u $USER) -v /data/blastdb:/blastdb /scratch:/scratch --name pathopred -dit nanjiang/pathopred

1. Now pathopred can be run. Some example files are provided in the container. To run pathopred for the sequence with identifier P26439 contained in the file P26439.fasta, which has any number of variants specified in file variants, and output predictions to the folder pathopred_output, the following command would be run

docker exec --user user pathopred script /dev/null -c "export HOME=/home/user; /app/pathopred/master_pathopred.sh /app/pathopred/example_files/P26439.fasta /app/pathopred/example_files/variants /scratch/pathopred_output P26439"

You can also specify the number of cores to use for BLAST with the -cpu option, so to use for example 2 cores, the following would be run

docker exec --user user pathopred script /dev/null -c "export HOME=/home/user; /app/pathopred/master_pathopred.sh /app/pathopred/example_files/P26439.fasta /app/pathopred/example_files/variants /scratch/pathopred_output P26439 -cpu 2"

1. The generated predictions can then be inspected in the file output_predictions in the output folder.

2. Input files are expected to be in a certain format. See the below section for details.

⁠Input format

The input arguments to the script are on the form

master_pathopred.sh <SEQ> <MUT> <OUTDIR> <ID>

SEQ is a fasta file containing your amino acid sequence. MUT is a file containing variants for the specified sequence. Individual variants are on the format of reference AA, position, altered AA, e.g. A509G. The variants should be specified below their corresponding sequence identifier starting with >. The file could look like the following:

>P26439
P186L
P222H
P222Q

OUTDIR is a directory to write output predictions to. ID is the identifier of the sequence, which currently has to be explicitly given, e.g. P26439 in this example.

Note that only uniprot identifiers are currently supported.

⁠Building on your own

Inside the folder containing the Dockerfile, run the following to build:

docker build -t pathopred .

⁠Running the container attached

You can run the container in attached mode using, for example:

docker run --name pathopred --rm -i -t -v <nr path>:/blastdb pathopred bash

where nr path is a path to a NCBI nr database.

⁠Using inside the container

Run a prediction for a sequence with any number of variants using

bash master_pathopred.sh <SEQ> <MUT> <OUTDIR> <ID>

Some example files are provided. You can do a test run using these, for example:

bash master_pathopred.sh example_files/P26439.fasta example_files/variants /home/user/output P26439