diff --git a/src/find_interaction_cluster/community_figures/__main__.py b/src/find_interaction_cluster/community_figures/__main__.py
index 259b1179f4760706674a387570e16faec9b5bc6b..884f616020ab508cd8d371331b4759b5773972c2 100644
--- a/src/find_interaction_cluster/community_figures/__main__.py
+++ b/src/find_interaction_cluster/community_figures/__main__.py
@@ -49,7 +49,7 @@ def load_and_check_table(table: str, feature: str, target_col: str):
return df
-@lp.parse(table="file", test_type=["lm", "permutation"],
+@lp.parse(table="file", test_type=["lm", "permutation", "logit"],
iteration="20 < iteration", sd_community = ["y", "n", "Y", "N"])
def create_community_figures(table: str, feature: str, target_col: str,
output: str, outfile: str, test_type: str,
diff --git a/src/find_interaction_cluster/community_figures/fig_functions.py b/src/find_interaction_cluster/community_figures/fig_functions.py
index a05c2561a6626fe7166b89b3593b1605bfdf321d..2c5080c12168edbba00bdec093f509b79d37f661 100644
--- a/src/find_interaction_cluster/community_figures/fig_functions.py
+++ b/src/find_interaction_cluster/community_figures/fig_functions.py
@@ -54,8 +54,8 @@ def get_community_table(communities: List[List[str]],
return pd.DataFrame(dic)
-def lm_maker_summary(df: pd.DataFrame, outfile: Path, target_col: str
- ) -> pd.DataFrame:
+def lm_maker_summary(df: pd.DataFrame, outfile: Path, target_col: str,
+ test_type: str) -> pd.DataFrame:
"""
Make the lm analysis to see if the exon regulated by a splicing factor \
are equally distributed among the communities.
@@ -66,15 +66,22 @@ def lm_maker_summary(df: pd.DataFrame, outfile: Path, target_col: str
and the size of the community of the feature if it has one (None, else).
:param outfile: A name of a file
:param target_col: The name of the column containing the data of interest
+ :param test_type: The type of test to make (permutation or lm)
:return: the pvalue of lm
"""
pandas2ri.activate()
+ if test_type == "lm":
+ mod = f"mod <- lm({target_col} ~ log(community_size) + " \
+ f"community, data=data)"
+ else:
+ mod = f"mod <- glm({target_col} ~ log(community_size) + community," \
+ f"data=data, family=binomial(link='logit'))"
lmf = r(
"""
require("DHARMa")
function(data, folder, partial_name) {
- mod <- lm(%s ~ log(community_size) + community, data=data)
+ %s
simulationOutput <- simulateResiduals(fittedModel = mod, n = 250)
png(paste0(folder, "/dignostics_summary", partial_name, ".png"))
plot(simulationOutput)
@@ -82,7 +89,7 @@ def lm_maker_summary(df: pd.DataFrame, outfile: Path, target_col: str
return(as.data.frame(summary(mod)$coefficients))
}
- """ % target_col)
+ """ % mod)
folder = outfile.parent / "diagnostics"
folder.mkdir(parents=True, exist_ok=True)
partial_name = outfile.name.replace('.pdf', '')
@@ -97,7 +104,7 @@ def lm_maker_summary(df: pd.DataFrame, outfile: Path, target_col: str
def lm_with_ctrl(df: pd.DataFrame,
- target_col: str, outfile: Path,
+ target_col: str, outfile: Path, test_type: str
) -> Tuple[pd.DataFrame, pd.DataFrame]:
"""
:param df: A dataframe containing the id of the chosen `feature` \
@@ -106,6 +113,7 @@ def lm_with_ctrl(df: pd.DataFrame,
and the size of the community of the feature if it has one (None, else).
:param target_col: The name of the column containing the data of interest
:param outfile: File that will contains the final figure
+ :param test_type: The type of test to make (permutation or lm)
:return: The dataframe with ctrl exon and \
The dataframe with the p-value compared to the control \
list of feature.
@@ -114,11 +122,12 @@ def lm_with_ctrl(df: pd.DataFrame,
size = df.loc[df["community"] == "C-CTRL", :].shape[0]
df['community_size'] = df['community_size'].fillna(size)
df['community_size'] = df['community_size'].astype(int)
- return df, lm_maker_summary(df, outfile, target_col)
+ return df, lm_maker_summary(df, outfile, target_col, test_type)
def expand_results_lm(df: pd.DataFrame, rdf: pd.DataFrame,
- target_col: str, feature: str) -> pd.DataFrame:
+ target_col: str, feature: str,
+ test_type: str) -> pd.DataFrame:
"""
Merge df and rdf together.
@@ -131,9 +140,10 @@ def expand_results_lm(df: pd.DataFrame, rdf: pd.DataFrame,
exons.
:param target_col: The name of the column containing the data of interest
:param feature: The feature of interest
+ :param test_type: The kind of test to make
:return: The merged dataframe: i.e df with the stats columns
"""
- p_col = "Pr(>|t|)"
+ p_col = "Pr(>|t|)" if test_type == "lm" else "Pr(>|z|)"
df = df[[f"id_{feature}", target_col, "community",
"community_size"]].copy()
rdf = rdf[["community", "community_size", p_col, target_col]].copy()
@@ -347,7 +357,7 @@ def display_size_fig(g: sns.FacetGrid, display_size: bool, target_col: str,
return g
-def make_barplot(df_bar: pd.DataFrame, outfile: Path,
+def make_barplot(df_bar: pd.DataFrame, outfile: Path, test_type: str,
target_col: str, feature: str, target_kind: str = "",
sd_community: Optional[str] = "sd",
display_size: bool = False) -> None:
@@ -360,6 +370,7 @@ def make_barplot(df_bar: pd.DataFrame, outfile: Path,
:param outfile: File were the figure will be stored
:param target_kind: An optional name that describe a bit further \
target_col.
+ :param test_type: The kind of test to perform
:param target_col: The name of the column containing the data of interest
:param feature: The king of feature of interest
:param sd_community: sd to display community error bar, None to display \
@@ -378,7 +389,7 @@ def make_barplot(df_bar: pd.DataFrame, outfile: Path,
g.fig.subplots_adjust(top=0.9)
target_kind = f" ({target_kind})" if target_kind else ""
g.fig.suptitle(f"Mean frequency of {target_col}{target_kind}"
- f"among community of {feature}s\n"
+ f" among community of {feature}s\n"
f"(stats obtained with a lm test)")
g = display_size_fig(g, display_size, target_col, df_bar)
g.ax.set_ylabel(f'Frequency of {target_col}')
@@ -439,7 +450,7 @@ def make_barplot_perm(df_bar: pd.DataFrame, outfile: Path,
g.fig.subplots_adjust(top=0.9)
target_kind = f" ({target_kind})" if target_kind else ""
g.fig.suptitle(f"Mean frequency of {target_col}{target_kind}"
- f"among community of {feature}s\n"
+ f" among community of {feature}s\n"
f"(stats obtained with a permutation test)")
g = display_size_fig(g, display_size, target_col, df_bar)
g.ax.set_ylabel(f'Frequency of {target_col}')
@@ -483,8 +494,8 @@ def barplot_creation(df_bar: pd.DataFrame, outfig: Path,
each one of them False to display nothing. (default False)
"""
sd_community = "sd" if sd_community else None
- if test_type == "lm":
- make_barplot(df_bar, outfig, cpnt, feature, target_kind,
+ if test_type != "permutation":
+ make_barplot(df_bar, outfig, test_type, cpnt, feature, target_kind,
display_size=display_size)
else:
make_barplot_perm(df_bar, outfig, cpnt, feature, target_kind,
@@ -553,11 +564,11 @@ def create_community_fig(df: pd.DataFrame, feature: str,
each one of them False to display nothing. (default False)
"""
if dic_com is None:
- dic_com = {} if test_type == 'lm' \
+ dic_com = {} if test_type != 'permutation' \
else get_feature_by_community(df, feature)
- if test_type == "lm":
- ndf, rdf = lm_with_ctrl(df, target_col, outfile_ctrl)
- df_bar = expand_results_lm(ndf, rdf, target_col, feature)
+ if test_type != "permutation":
+ ndf, rdf = lm_with_ctrl(df, target_col, outfile_ctrl, test_type)
+ df_bar = expand_results_lm(ndf, rdf, target_col, feature, test_type)
else:
rdf = perm_with_ctrl(df, feature, target_col, dic_com, iteration)
df_bar = expand_results_perm(df, rdf, target_col, feature, iteration)