phynteny

#!/usr/bin/env python

"""
Phynteny: synteny-based annotation of phage genes
"""

import datetime
import sys
import time

import click
import pkg_resources
from loguru import logger

from phynteny_utils import format_data, handle_genbank, predictor

__author__ = "Susanna Grigson"
__maintainer__ = "Susanna Grigson"
__license__ = "MIT"
__version__ = "0"
__email__ = "susie.grigson@gmail.com"
__status__ = "development"


@click.command()
@click.argument("infile", type=click.Path(exists=True))
@click.option(
    "-o",
    "--out",
    type=click.STRING,
    default="",
    help="output directory",
)
@click.option("-f", "--force", is_flag=True, help="Overwrite output directory")
@click.option(
    "-m",
    "--models",
    type=click.Path(exists=True),
    help="Path to directory containing phynteny models",
    default=pkg_resources.resource_filename("phynteny_utils", "models"),
)
@click.option(
    "-c",
    "--confidence_path",
    type=click.Path(exists=True),
    help="Dictionary of kernel desnity estimators to use for predicting confidence",
    default=pkg_resources.resource_filename(
        "phynteny_utils", "phrog_annotation_info/confidence_kde.pkl"
    ),
)
@click.version_option(version=__version__)
def main(infile, out, force, models, confidence_path):
    """
    Phynteny: synteny-based annotation of phage genes
    """

    # get the start time
    start_time = time.time()

    # generate the output directory
    format_data.instantiate_dir(out, force)

    # generate the logging object
    logger.add(out + "/phynteny.log", level="DEBUG")
    logger.info("Starting Phynteny")

    # get the absolute paths to phrog annotation files, model and confidence_kde
    phrog_categories = pkg_resources.resource_filename(
        "phynteny_utils", "phrog_annotation_info/phrog_integer.pkl"
    )
    logger.info(f"PHROG integer information located at: {phrog_categories}")
    category_names = pkg_resources.resource_filename(
        "phynteny_utils", "phrog_annotation_info/integer_category.pkl"
    )
    logger.info(f"PHROG category information located at: {category_names}")
    categories = format_data.get_dict(category_names)
    phrog_integer = format_data.get_dict(phrog_categories)

    # get entries in the genbank file
    logger.info("Reading genbank file!")
    gb_dict = handle_genbank.get_genbank(infile)
    if not gb_dict:
        click.echo("Error: no sequences found in genbank file")
        logger.critcal("No sequences found in genbank file. Nothing to annotate")
        sys.exit()

    # create predictor object
    gene_predictor = predictor.Predictor(
        models, phrog_categories, confidence_path, category_names
    )

    # generate predictions
    logger.info(f"Confidence object located at {confidence_path}")
    genbank_file = out + "/phynteny.gbk"
    phynteny_dict = predictor.run_phynteny(
        genbank_file, gene_predictor, gb_dict, categories
    )
    logger.info(f"Finished predicting. Genbank file located at {genbank_file}")

    # output to a table
    logger.info("Generating table...")
    table_file = out + "/phynteny.tsv"
    found = predictor.generate_table(
        table_file, phynteny_dict, categories, phrog_integer
    )
    logger.info(f"Generated table. Table located at {table_file}")
    logger.info(
        "Phynteny was able to add annotations for "
        + str(found)
        + " genes with a confidence of at least 90%"
    )
    logger.info("Done :)")

    # show the time elapsed
    elapsed_time = round(time.time() - start_time, 2)
    logger.info(f"Elpased time: {elapsed_time} seconds")


if __name__ == "__main__":
    main()