From cc536f0beda6c23e3ccc326b06a54a8277fc3e27 Mon Sep 17 00:00:00 2001
From: Fontrodona Nicolas <nicolas.fontrodona@ens-lyon.fr>
Date: Wed, 27 Jan 2021 11:24:52 +0100
Subject: [PATCH] src/find_interaction_cluster/clip_figures/__main__.py
src/find_interaction_cluster/clip_figures/clip_analyser.py: ad dinflation
parameter
---
.../clip_figures/__main__.py | 4 ++-
.../clip_figures/clip_analyser.py | 33 +++++++++++--------
2 files changed, 23 insertions(+), 14 deletions(-)
diff --git a/src/find_interaction_cluster/clip_figures/__main__.py b/src/find_interaction_cluster/clip_figures/__main__.py
index e3c9fffc..5c07cf85 100644
--- a/src/find_interaction_cluster/clip_figures/__main__.py
+++ b/src/find_interaction_cluster/clip_figures/__main__.py
@@ -15,6 +15,7 @@ from .clip_analyser import clip_folder_analysis
@lp.parse(test_type=["permutation", "lm"], feature=["gene", "exon"])
def clip_analysis(clip_folder: str, weight: int = -1,
global_weight: int = -1, same_gene: bool = True,
+ inflation: float = 1.5,
feature: str = "exon",
project: str = "GSM1018963_GSM1018964",
test_type: str = "permutation", iteration: int = 10000,
@@ -32,6 +33,7 @@ def clip_analysis(clip_folder: str, weight: int = -1,
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 (default 1.5)
:param feature: The feature we want to analyse (default 'exon')
:param clip_folder: A folder containing clip file
:param test_type: The kind of test to perform for frequency analysis. \
@@ -57,7 +59,7 @@ def clip_analysis(clip_folder: str, weight: int = -1,
if not clip_folder.is_dir():
raise NotADirectoryError(f"{clip_folder} is not an existing directory")
clip_folder_analysis(clip_folder, project, weight, global_weight,
- same_gene, feature, test_type, iteration,
+ same_gene, inflation, feature, test_type, iteration,
display_size, community_file, sl_reg, ps,
logging_level)
diff --git a/src/find_interaction_cluster/clip_figures/clip_analyser.py b/src/find_interaction_cluster/clip_figures/clip_analyser.py
index 5e53c78c..eb2c8aae 100644
--- a/src/find_interaction_cluster/clip_figures/clip_analyser.py
+++ b/src/find_interaction_cluster/clip_figures/clip_analyser.py
@@ -60,7 +60,8 @@ def bedtools_intersect(gene_bed: Path, clip_bed: Path,
def find_or_create_community(project: str, weight: int, global_weight: int,
- same_gene: bool, feature: str) -> Path:
+ same_gene: bool, inflation: float,
+ feature: str) -> Path:
"""
Find the community file of interest, or if it doesn't exist, create it.
@@ -72,24 +73,26 @@ def find_or_create_community(project: str, weight: int, global_weight: int,
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 feature: The feature we want to analyse (default 'exon')
:return: The file containing community
- >>> res = find_or_create_community("", 3, 3, True, "gene")
+ >>> res = find_or_create_community("", 3, 3, True, 1.5, "gene")
>>> res_path = str(res.relative_to(Path(__file__).parents[3]))
>>> res_path.endswith("global-weight-3_weight-3_same_gene-True_gene.txt")
True
>>> f = "results/community_of_co-localized-exons/" + \
- "communities/weight-3_global_weight-3"
+ "communities/weight-3_global_weight-3-1.5"
>>> res_path.startswith(f)
True
"""
cfile = ConfigGraph.get_community_file(project, weight, global_weight,
- same_gene, feature, ".txt")
+ same_gene, inflation, feature,
+ ".txt")
if not cfile.is_file():
project = get_projects(global_weight, project)
multiple_community_launcher(weight, global_weight, project, same_gene,
- False, feature)
+ inflation, False, feature)
if not cfile.is_file():
raise FileNotFoundError(f"The file {cfile} could not be created !")
return cfile
@@ -176,7 +179,7 @@ def create_table(feature: str, clip_file: Path,
:param com_file: A file containing communities
:return: The final dataframe that can be used to create clip figures
- >>> cf = find_or_create_community("", 3, 3, True, "gene")
+ >>> cf = find_or_create_community("", 3, 3, True, 1.5, "gene")
>>> create_table("gene", ConfigClip.test_clip_bed,
... ConfigClip.test_gene_bed, cf)
id_gene clip_peak peak_density community community_size
@@ -203,7 +206,7 @@ def create_table(feature: str, clip_file: Path,
def select_community_file(project: str, weight: int, global_weight: int,
- same_gene: bool, feature: str,
+ same_gene: bool, inflation: float, feature: str,
community_file: str = "") -> Tuple[Path, Path]:
"""
Return the community file and output folder that will be used.
@@ -216,6 +219,7 @@ def select_community_file(project: str, weight: int, global_weight: int,
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 feature: The feature we want to analyse (default 'exon')
:param community_file: A file containing custom communities. If \
it equals to '' then weight, global weight and same genes parameter are \
@@ -225,7 +229,7 @@ def select_community_file(project: str, weight: int, global_weight: int,
"""
if community_file == "":
com_file = find_or_create_community(project, weight, global_weight,
- same_gene, feature)
+ same_gene, inflation, feature)
output = com_file.parent / f"CLIP_community_figures_{feature}"
else:
com_file = Path(community_file)
@@ -297,7 +301,8 @@ def add_regulation_column(df_table: pd.DataFrame, sf_name: str, feature: str,
def create_figure(project: str, weight: int, global_weight: int,
- same_gene: bool, feature: str, clip_file: Path,
+ same_gene: bool, inflation: float, feature: str,
+ clip_file: Path,
feature_bed: Path, test_type: str = "permutation",
iteration: int = 10000, display_size: bool = False,
community_file: str = "", sl_reg: bool = False) -> None:
@@ -311,6 +316,7 @@ def create_figure(project: str, weight: int, global_weight: int,
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 feature: The feature we want to analyse (default 'exon')
:param clip_file: A bed file containing clip
:param feature_bed: A bed files containing exons or genes depending on \
@@ -328,7 +334,7 @@ def create_figure(project: str, weight: int, global_weight: int,
"""
logging.info(f"Working on {clip_file}")
com_file, output = select_community_file(project, weight, global_weight,
- same_gene, feature,
+ same_gene, inflation, feature,
community_file)
output.mkdir(exist_ok=True, parents=True)
outfile = output / f"{clip_file.name.split('.')[0]}.pdf"
@@ -343,8 +349,8 @@ def create_figure(project: str, weight: int, global_weight: int,
def clip_folder_analysis(clip_folder: Path, project: str, weight: int,
- global_weight: int, same_gene: bool, feature: str,
- test_type: str = "permutation",
+ global_weight: int, same_gene: bool, inflation: float,
+ feature: str, test_type: str = "permutation",
iteration: int = 10000, display_size: bool=False,
community_file: str = "", sl_reg: bool = False,
ps: int = 1, logging_level: str = "DEBUG") -> None:
@@ -358,6 +364,7 @@ def clip_folder_analysis(clip_folder: Path, project: str, weight: int,
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 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. \
@@ -381,7 +388,7 @@ def clip_folder_analysis(clip_folder: Path, project: str, weight: int,
pool = mp.Pool(processes=min(len(files), ps))
processes = []
for mfile in files:
- args = [project, weight, global_weight, same_gene, feature,
+ args = [project, weight, global_weight, same_gene, inflation, feature,
mfile, feature_bed, test_type, iteration, display_size,
community_file, sl_reg]
processes.append(pool.apply_async(create_figure, args))
--
GitLab