parallel_assign_taxonomy_blast.py – Parallel taxonomy assignment using BLAST
Description:
This script performs like the assign_taxonomy.py script, but is intended to make use of multicore/multiprocessor environments to perform analyses in parallel.
Usage: parallel_assign_taxonomy_blast.py [options]
Input Arguments:
Note
[REQUIRED]
- -i, --input_fasta_fp
- Full path to input_fasta_fp [REQUIRED]
- -t, --id_to_taxonomy_fp
- Full path to id_to_taxonomy mapping file [REQUIRED]
- -o, --output_dir
- Full path to store output files [REQUIRED]
[OPTIONAL]
- -r, --reference_seqs_fp
- Ref seqs to blast against. Must provide either –blast_db or –reference_seqs_db for assignment with blast [default: None]
- -b, --blast_db
- Database to blast against. Must provide either –blast_db or –reference_seqs_db for assignment with blast [default: None]
- -e, --e_value
- Maximum e-value to record an assignment, only used for blast method [default: 0.001]
- -B, --blastmat_dir
- Full path to directory containing blastmat file [default: /Applications/blast-2.2.22/data/]
- -N, --assign_taxonomy_fp
- Full path to scripts/assign_taxonomy.py [default: /Users/caporaso/code/Qiime/scripts/assign_taxonomy.py]
- -O, --jobs_to_start
- Number of jobs to start [default: 2]
- -P, --poller_fp
- Full path to qiime/parallel/poller.py [default: /Users/caporaso/code/Qiime/scripts/poller.py]
- -R, --retain_temp_files
- Retain temporary files after runs complete (useful for debugging) [default: False]
- -S, --suppress_submit_jobs
- Only split input and write commands file - don’t submit jobs [default: False]
- -T, --poll_directly
- Poll directly for job completion rather than running poller as a separate job. If -T is specified this script will not return until all jobs have completed. [default: False]
- -U, --cluster_jobs_fp
- Path to cluster jobs script (defined in qiime_config) [default: /Users/caporaso/code/Qiime/scripts/start_parallel_jobs.py]
- -W, --suppress_polling
- Suppress polling of jobs and merging of results upon completion [default: False]
- -X, --job_prefix
- Job prefix [default: descriptive prefix + random chars]
- -Y, --python_exe_fp
- Full path to python executable [default: python]
- -Z, --seconds_to_sleep
- Number of seconds to sleep between checks for run completion when polling runs [default: 6]
Output:
Mapping of sequence identifiers to taxonomy and quality scores.
Example:
Assign taxonomy to all sequences in the input file (-i) via five (-O) independent jobs using BLAST with the id to taxonomy mapping file (-t) and reference sequence template file (-r), and write the results (-o) to /home/qiime_user/out/. BE SURE TO SPECIFY FULL PATHS!
parallel_assign_taxonomy_blast.py -O 5 -i /home/qiime_user/inseqs.fasta -t /home/qiime_user/at_id_to_taxonomy.txt -r /home/qiime_user/at_refseqs.fasta -o /home/qiime_user/out/