src/find_interaction_cluster/clip_figures/clip_launcher_4_many_communities.py:...

src/find_interaction_cluster/clip_figures/clip_launcher_4_many_communities.py: create for each clip the community figures obtained using many community files

src/find_interaction_cluster/clip_figures/clip_launcher_4_many_communities.py

+#!/usr/bin/env python3
+
+# -*- coding: UTF-8 -*-
+
+"""
+Description: Launch Clip analysis for many communities
+"""
+
+
+import logging
+from typing import Tuple, Dict
+from pathlib import Path
+from .config import ConfigClip
+from .clip_analyser import create_table, \
+    add_regulation_column, create_community_fig, find_or_create_community, \
+    ConfigGraph, logging_def
+import multiprocessing as mp
+from itertools import product
+import lazyparser as lp
+import numpy as np
+import subprocess as sp
+
+
+def select_community_file(project: str, weight: int, global_weight: int,
+                          same_gene: bool, inflation: float, cell_line: str,
+                          feature: str, clip_file: Path,
+                          community_file: str = "") -> Tuple[Path, Path]:
+    """
+    Return the community file and output folder that will be used.
+
+    :param project: The name of the project of interest
+    :param weight: The minimum weight of interaction to consider
+    :param global_weight: The global weight to consider. if \
+    the global weight is equal to 0 then then density figure are calculated \
+    by project, else all projet are merge together and the interaction \
+    seen in `global_weight` project are taken into account
+    :param same_gene: Say if we consider as co-localised, exons within the \
+    same gene (True) or not (False) (default False)
+    :param inflation: The inflation parameter
+    :param cell_line: Interactions are only selected from projects made on \
+    a specific cell line (ALL to disable this filter). (default ALL)
+    :param feature: The feature we want to analyse (default 'exon')
+    :param clip_file: A bed file containing clip
+    :param community_file: A file containing custom communities. If \
+    it equals to '' then weight, global weight and same genes parameter are \
+    used to find the community files computed with ChIA-PET data.
+    :return: The community file used and the output folder used.
+
+    """
+    if community_file == "":
+        com_file = find_or_create_community(project, weight, global_weight,
+                                            same_gene, inflation, cell_line,
+                                            feature)
+    else:
+        com_file = Path(community_file)
+        if not com_file.is_file():
+            raise FileNotFoundError(f"File {com_file} was not found !")
+    outname = clip_file.name.split('.')[0]
+    output = ConfigClip.output_folder / \
+        f"multiTAD_CLIP_community_figures-{feature}-{cell_line}" / outname
+    return com_file, output
+
+
+def create_figure(p: Dict , clip_file: Path,
+                  feature_bed: Path,
+                  community_file: Tuple[str, str],
+                  test_type: str = "permutation",
+                  iteration: int = 10000, display_size: bool = False,
+                  sl_reg: bool = False) -> Path:
+    """
+    Create the final figure
+    :param p: A dictionary containing parameter used to compute \
+    HipMCL communities
+    :param clip_file: A bed file containing clip
+    :param feature_bed: A bed files containing exons or genes depending on \
+    feature parameter.
+    :param test_type: The king of test to perform for frequency analysis. \
+    (default 'lm') (choose from 'lm', 'permutation')
+    :param iteration: The number of iteration to make
+    :param community_file: A Tuple containing a  file containing custom \
+    communities. If it equals to '' then weight, global weight and \
+    same genes parameter are used to find the community files computed \
+     with ChIA-PET data. The second item of the tuple is it's name.
+    :param display_size: True to display the size of the community. \
+    False to display nothing. (default False)
+    :param sl_reg: True to display the FaRLine regulation of the \
+    same factor, False to not display it.
+    :return: Folder containing he figures
+    """
+    logging.info(f"Working on {clip_file} - {community_file[0]} - "
+                 f"{community_file[1]}")
+    com_file, output = select_community_file(p["project"], p["weight"],
+                                             p["global_weight"],
+                                             p["same_gene"], p["inflation"],
+                                             p["cell_line"],
+                                             p["feature"], clip_file,
+                                             community_file[0])
+    output.mkdir(exist_ok=True, parents=True)
+    outfile = output / f"{community_file[1]}.tmp.pdf"
+    final_table = create_table(p["feature"], clip_file, feature_bed, com_file)
+    if sl_reg:
+        final_table = add_regulation_column(final_table,
+                                            clip_file.name.split("_")[0],
+                                            p["feature"])
+    create_community_fig(final_table, p["feature"], "peak_density", outfile,
+                         test_type, iteration=iteration,
+                         display_size=display_size)
+    return outfile.parent
+
+
+def merge_figures(folder: Path) -> None:
+    """
+    Merge the figures together using imageMagick
+
+    :param folder: A folder containing pdf files
+    """
+    fig_name = folder.name
+    cmd = f"montage -geometry +1+1 -tile 1X6  " \
+          f"-compress jpeg -density 100 " \
+          f"{folder}/*.tmp.pdf {folder}/{fig_name}.pdf"
+    sp.check_call(cmd, shell=True)
+
+
+@lp.parse(test_type=["permutation", "lm"], feature=["gene", "exon"])
+def clip_folder_analysis(clip_folder: str, weight: int,
+                         global_weight: int, same_gene: bool = True,
+                         project: str = "GSM1018963_GSM1018964",
+                         inflation: float = 1.5,
+                         cell_line: str = "ALL",
+                         feature: str = "exon",
+                         test_type: str = "permutation",
+                         iteration: int = 10000, display_size: bool=False,
+                         sl_reg: bool = False) -> None:
+    """
+    Create the final figure
+    :param project: The name of the project of interest
+    :param weight: The minimum weight of interaction to consider
+    :param global_weight: The global weight to consider. if \
+    the global weight is equal to 0 then then density figure are calculated \
+    by project, else all projet are merge together and the interaction \
+    seen in `global_weight` project are taken into account
+    :param same_gene: Say if we consider as co-localised, exons within the \
+    same gene (True) or not (False) (default False)
+    :param inflation: The inflation parameter
+    :param cell_line: Interactions are only selected from projects made on \
+    a specific cell line (ALL to disable this filter). (default ALL)
+    :param feature: The feature we want to analyse (default 'exon')
+    :param clip_folder: A folder containing clip file
+    :param test_type: The king of test to perform for frequency analysis. \
+    (default 'lm') (choose from 'lm', 'permutation')
+    :param iteration: The number of iteration to make
+    :param display_size: True to display the size of the community. \
+    False to display nothing. (default False)
+    :param sl_reg: True to display the FaRLine regulation of the \
+    same factor, False to not display it.
+    :param ps: The number of processes to create (default 1)
+    """
+    logging_def(ConfigGraph.community_folder, __file__, "INFO")
+    clip_folder = Path(clip_folder)
+    feature_bed = ConfigClip.bed_gene if feature == "gene" \
+        else ConfigClip.bed_exon
+    files = list(clip_folder.glob("*.bed")) + \
+        list(clip_folder.glob("*.bed.gz"))
+    files = [files[0]]
+
+    processes = []
+    p = {"project": project, "weight": weight, "global_weight": global_weight,
+         "same_gene": same_gene, "inflation": inflation,
+         "cell_line": cell_line, "feature": feature}
+
+    prod = list(product(files, zip(ConfigClip.communities,
+                           ConfigClip.communities_name)))
+    pool = mp.Pool(processes=min(len(prod), ConfigGraph.cpu))
+    for mfile,  community_file in prod:
+        if community_file[0] == "":
+            tmp = \
+                f"HIPMCL_g{global_weight}_w{weight}_{inflation}"
+            community_file = (community_file[0], tmp)
+        args = [p, mfile, feature_bed, community_file, test_type, iteration,
+                display_size, sl_reg]
+        processes.append(pool.apply_async(create_figure, args))
+    list_path = [str(p.get(timeout=None)) for p in processes]
+    list_path = np.unique(list_path)
+    for my_folder in list_path:
+        merge_figures(Path(my_folder))
+
+
+if __name__ == "__main__":
+    clip_folder_analysis()
\ No newline at end of file