This repository contains the update of deepARG (deep learning based approach for antibiotic resistance gene annotation)
DeepARG+ has been released in a docker image to avoid library issues and conflict with newer versions of the libraries.
docker run --runtime=nvidia -it -v $PWD:/data/ --rm gaarangoa/deepargplus:latest deepARG+ train \
--inputdir /data/ \
--outdir /data/ \
--prefix DL \
--epoch 10 \
--batch 32
deepARG+ train \
--inputdir ./ \
--outdir ./ \
--prefix DL \
--epoch 2 \
--batch 32
docker run --runtime=nvidia -it -v $PWD:/data/ --rm gaarangoa/deepargplus:latest deepARG+ predict \
--inputfile /data/tests/b.fasta \
--wordvec-model /data/wvecmodel/model.bin \
--deeparg-model /data/DL.001.hdf5 \
--deeparg-parameters /data/DL.parameters.json \
--outdir /data/tests/ \
--prefix bla
deepARG+ predict --inputfile b.fa --wordvec-model ../wvecmodel/model.bin --deeparg-model ../DL.001.hdf5 --deeparg-parameters ../DL.parameters.json --outdir ./ --prefix bla
You need to make sure that your fasta file header follows this schema:
>gene_id|arg_category|arg_name|arg_group
arg_name refers to the name of the arg e.g., OXA-1 arg_group refers to the grouping of very similar args, for instance OXA
First convert the fasta file to a word vector representation
deepARG+ fasta2vec --help
Check the log file to make sure the script ran without problems.
Second, run the training
deepARG+ train --help
Thus, the model file will be generated
Content type
Image
Digest
Size
1.7 GB
Last updated
over 7 years ago
docker pull gaarangoa/deepargplus