pangeblocks_docker

#!/usr/bin/env python3
"""
Compute a Variation Graph from an MSA.

This programs only parses some command line options and produces the snakemake call to compute a single graph.
"""
import argparse
import pathlib
import subprocess
import os

# Command line options
parser = argparse.ArgumentParser()
parser.add_argument(
    "--path-msa",
    help="path to MSA in .fa format",
    dest="path_msa",
    type=pathlib.Path,
    default="/data/sars-cov-2.fa",
)
parser.add_argument(
    "--path-gfa",
    help="path to save the output in GFA format",
    dest="path_gfa",
    type=pathlib.Path,
    default="/results/sars-cov-2.gfa",
)
parser.add_argument(
    "--dir-subsolutions",
    help="directory where suboptimal coverage are saved",
    dest="dir_subsolutions",
    type=pathlib.Path,
    default="/results/submsa",
)
parser.add_argument(
    "--path-vert-blocks",
    help="json file with vertical blocks",
    dest="path_vert_blocks",
    type=pathlib.Path,
)
parser.add_argument(
    "--log-level",
    default="ERROR",
    dest="log_level",
    help="set log level (ERROR/WARNING/INFO/DEBUG)",
)
parser.add_argument(
    "--obj-function",
    help="the objective function to optimize",
    dest="obj_function",
    choices=["nodes", "strings", "weighted", "depth", "depth_and_len"],
    default="weighted",
)
parser.add_argument(
    "--penalization",
    help="used only with the weighted obj function",
    dest="penalization",
    default=1000,
    type=int,
)
parser.add_argument(
    "--min-len",
    help="used only with the weighted obj function",
    dest="min_len",
    default=2,
    type=int,
)
parser.add_argument(
    "--time-limit", help="Timeout (in minutes)", dest="time_limit", default=10, type=int
)
parser.add_argument(
    "--threshold-vertical-blocks",
    help="minimum width of a vertical block",
    dest="alpha",
    default=50,
    type=int,
)
parser.add_argument(
    "--min-coverage", help="", dest="min_coverage", default=0.2, type=float
)
parser.add_argument(
    "--larger-decomposition",
    help="",
    dest="standard_decomposition",
    action="store_true",
)
parser.add_argument(
    "--consistent",
    help="use an alpha-consistent strategy",
    dest="consistent",
    action="store_true",
)
parser.add_argument(
    "--cores", help="Number of cores to be used", dest="threads", default="2", type=int
)
parser.add_argument(
    "--submsa_threads", help="", dest="submsa_threads", default=2, type=int
)
parser.add_argument("--ilp-threads", help="", dest="ilp_threads", default=2, type=int)
parser.add_argument(
    "--max-memory",
    help="Maximum RAM used (in MBytes)",
    dest="max_memory",
    default=5120,
    type=int,
)
parser.add_argument(
    "--min-rows-block", help="", dest="min_rows_block", default=0, type=int
)
parser.add_argument(
    "--max-rows-block", help="", dest="max_rows_block", default=0, type=int
)
parser.add_argument(
    "--max-msa-size", help="", dest="max_msa_size", default=80000, type=int
)
args = parser.parse_args()

cmds = [
    f"snakemake",
    f"-c{args.threads}",
    f"--printshellcmds",
    f"-s",
    "pangeblocks_docker.smk",
    f"--directory={pathlib.Path(os.path.realpath(args.path_gfa)).parent}",
    f"--config",
    f"root_dir={pathlib.Path(os.path.realpath(__file__)).parents[0]}",
    f"PATH_MSA={args.path_msa}",
    f"PATH_OUTPUT={args.path_gfa}",
    f"OBJECTIVE_FUNCTION={args.obj_function}",
    f"PENALIZATION={args.penalization}",
    f"MIN_LEN={args.min_len}",
    f"TIME_LIMIT={args.time_limit}",
    f"THRESHOLD_VERTICAL_BLOCKS={args.alpha}",
    f"MIN_COVERAGE={args.min_coverage}",
    f"STANDARD={args.standard_decomposition}",
    f"ALPHA_CONSISTENT={args.consistent}",
    f"SUBMSAS={args.submsa_threads}",
    f"ILP={args.ilp_threads}",
    f"MEM_MB={args.max_memory}",
    f"MIN_ROWS_FIX_BLOCK={args.min_rows_block}",
    f"MIN_COLS_FIX_BLOCK={args.max_rows_block}",
    f"MAX_POSITIONS_SUBMSAS={args.max_msa_size}",
    f"LOG_LEVEL={args.log_level}",
]

print(" ".join(cmds))
subprocess.run(cmds)