diff --git a/src/find_interaction_cluster/clip_figures/clip_compo_analyser.py b/src/find_interaction_cluster/clip_figures/clip_compo_analyser.py
index 799e398a9694bc0895d77b20dc0aca33b10c90cb..408d8ffed1a99f38e4e00795a37b43ba7fa256da 100644
--- a/src/find_interaction_cluster/clip_figures/clip_compo_analyser.py
+++ b/src/find_interaction_cluster/clip_figures/clip_compo_analyser.py
@@ -363,11 +363,34 @@ def make_barplot(df_comp: pd.DataFrame, outfile: Path, cpnt: str,
plt.close()
+def update_composition_group(df_comp: pd.DataFrame, display_size: bool
+ ) -> pd.DataFrame:
+ """
+ Update the group name of the dataframe df_com.
+
+ :param df_comp: The dataframe containing the frequency of every `cpnt_type`
+ :param display_size: True to display the size of \
+ the community. False to display nothing.
+ :return: if display_size is False return df_comp else return df_comp \
+ with the group column containing the size of the groups.
+ """
+ if not display_size:
+ return df_comp
+ d = {
+ gr: f"{gr}({df_comp[df_comp['group'] == gr].shape[0]})"
+ for gr in df_comp["group"].unique()
+ }
+
+ df_comp["group"].map(d)
+ return df_comp
+
+
def create_figure_4_clip_table(clip_table_file: Path, feature: str,
threshold_feature: int, default_groups: int,
df_comp: pd.DataFrame, cpnt_type: str,
cpnt: str, df_id_ctrl: pd.DataFrame,
- write_comp_table: bool) -> None:
+ write_comp_table: bool,
+ display_size: bool) -> None:
"""
Create a component figure for a clip file
@@ -384,19 +407,21 @@ def create_figure_4_clip_table(clip_table_file: Path, feature: str,
:param cpnt: The kind of component of interest
:param df_id_ctrl: A dataframe indicating control exons
:param write_comp_table: True to save comp table, false else
+ :param display_size: True to display the size of the community .
+ False to display nothing. (default False)
"""
logging.info(f"Working on {clip_table_file.name} - {cpnt} ({cpnt_type})")
df_group = get_interest_groups(clip_table_file, feature,
threshold_feature, default_groups)
if df_group is None:
return None
+ df_group = df_group.drop_duplicates()
if feature == "gene":
df_group["id_gene"] = df_group["id_gene"].astype(int)
df_comp["id_gene"] = df_comp["id_gene"].astype(int)
df_group = pd.concat([df_group, df_id_ctrl], ignore_index=True)
-
if len(np.unique(df_group[f"id_{feature}"].values)) != df_group.shape[0]:
-
+ print(df_group[df_group.duplicated(subset=f"id_{feature}")])
raise ValueError("Found duplicates value in df_group, exiting...")
df_comp = df_comp.merge(df_group, how="left", on=f"id_{feature}")
df_comp = df_comp[-df_comp["group"].isna()]
@@ -408,6 +433,7 @@ def create_figure_4_clip_table(clip_table_file: Path, feature: str,
outfiles[0].parent.mkdir(exist_ok=True)
df_stat = create_stat_df(df_comp, cpnt)
df_stat.to_csv(outfiles[0], sep="\t", index=False)
+ df_comp = update_composition_group(df_comp, display_size)
make_barplot(df_comp, outfiles[1], cpnt, cpnt_type, sf_name)
if write_comp_table:
df_comp.to_csv(outfiles[2], sep="\t", index=False)
@@ -416,7 +442,7 @@ def create_figure_4_clip_table(clip_table_file: Path, feature: str,
def create_multiple_figures(clip_result_folder: Path, feature: str,
threshold_feature: int, default_groups: int,
cpnt_type: str, region: str = "gene",
- ps: int = 1,
+ ps: int = 1, display_size: bool = False,
logging_level: str = "DEBUG") -> None:
"""
Create multiple composition figures from a result clip folder.
@@ -433,6 +459,8 @@ def create_multiple_figures(clip_result_folder: Path, feature: str,
:param region: Only used when feature is gene. Corresponds to \
the region studied within genes.
:param ps: The number of processes to create
+ :param display_size: True to display the size of the community,
+ False to display nothing. (default False)
:param logging_level: The level of data to display (default 'DISABLE')
"""
logging_def(ConfigGraph.community_folder, __file__, logging_level)
@@ -453,7 +481,8 @@ def create_multiple_figures(clip_result_folder: Path, feature: str,
write_bar_table = True
for clip_file, cpnt in conditions:
args = [clip_file, feature, threshold_feature, default_groups,
- df_comp, cpnt_type, cpnt, df_ctrl, write_bar_table]
+ df_comp, cpnt_type, cpnt, df_ctrl, write_bar_table,
+ display_size]
write_bar_table = False
processes.append(pool.apply_async(create_figure_4_clip_table, args))
[p.get(timeout=None) for p in processes]
diff --git a/src/find_interaction_cluster/clip_figures/clip_compo_launcher.py b/src/find_interaction_cluster/clip_figures/clip_compo_launcher.py
index 8d4d05ed120b519f7f8fe60a0616b31ec86170c7..9edd36867c6e9cf1e17d8538a3df47bf6e829822 100644
--- a/src/find_interaction_cluster/clip_figures/clip_compo_launcher.py
+++ b/src/find_interaction_cluster/clip_figures/clip_compo_launcher.py
@@ -19,7 +19,8 @@ import multiprocessing as mp
def launcher(clip_result_folder: str, feature: str,
threshold_feature: int, default_groups: int,
cpnt_type: str, region: str = "gene",
- ps: int = 1, logging_level: str = "DISABLE") -> None:
+ ps: int = 1, display_size: bool=False,
+ logging_level: str = "DISABLE") -> None:
"""
Create multiple composition figures from a result clip folder.
@@ -35,6 +36,8 @@ def launcher(clip_result_folder: str, feature: str,
:param region: Only used when feature is gene. Corresponds to \
the region studied within genes. (default: 'gene')
:param ps: The number of processes to create (default 1)
+ :param display_size: True to display the size of the community. \
+ False to display nothing. (default False)
:param logging_level: The level of data to display (default 'DISABLE')
"""
clip_result_folder = Path(clip_result_folder)
@@ -42,7 +45,7 @@ def launcher(clip_result_folder: str, feature: str,
raise NotADirectoryError(f"Directory {clip_result_folder} don't exist")
create_multiple_figures(clip_result_folder, feature, threshold_feature,
default_groups, cpnt_type, region, ps,
- logging_level)
+ display_size, logging_level)
-
-launcher()
+if __name__ == "__main__":
+ launcher()