src/bed_handler/select_regulated_near_ctcf_exons.py: exon containing CTCF...

src/bed_handler/select_regulated_near_ctcf_exons.py: exon containing CTCF sites are dropped when the threshold is above 0

src/bed_handler/select_regulated_near_ctcf_exons.py

@@ -32,23 +32,19 @@ def filter_ctcf_distance_table(df: pd.DataFrame, reg: str, threshold: int,
     0    DSC2_1    -4  siPP_DOWN  1_1
     1    DSC2_2    -3  siPP_DOWN  1_2
     2    DSC2_3    -2  siPP_DOWN  1_3
-    4    DSC2_5     0  siPP_DOWN  1_5
     >>> rdf = filter_ctcf_distance_table(cdf, 'down', 2, 'upstream')
     >>> rdf[['exon_name', 'dist', 'group', 'id']]
       exon_name  dist      group   id
     2    DSC2_3    -2  siPP_DOWN  1_3
-    4    DSC2_5     0  siPP_DOWN  1_5
     >>> rdf = filter_ctcf_distance_table(cdf, 'down', 2, 'downstream')
     >>> rdf[['exon_name', 'dist', 'group', 'id']]
       exon_name  dist      group   id
-    4    DSC2_5     0  siPP_DOWN  1_5
     5    DSC2_6     1  siPP_DOWN  1_6
     6    DSC2_7     2  siPP_DOWN  1_7
     >>> rdf = filter_ctcf_distance_table(cdf, 'down', 2, 'both')
     >>> rdf[['exon_name', 'dist', 'group', 'id']]
       exon_name  dist      group   id
     2    DSC2_3    -2  siPP_DOWN  1_3
-    4    DSC2_5     0  siPP_DOWN  1_5
     5    DSC2_6     1  siPP_DOWN  1_6
     6    DSC2_7     2  siPP_DOWN  1_7
     >>> rdf = filter_ctcf_distance_table(cdf, 'up', 2, 'both')
@@ -65,7 +61,6 @@ def filter_ctcf_distance_table(df: pd.DataFrame, reg: str, threshold: int,
       exon_name  dist      group   id
     2    DSC2_3    -2  siPP_DOWN  1_3
     3    DSC2_4    -1    siPP_UP  1_4
-    4    DSC2_5     0  siPP_DOWN  1_5
     5    DSC2_6     1  siPP_DOWN  1_6
     6    DSC2_7     2  siPP_DOWN  1_7
     >>> filter_ctcf_distance_table(cdf, 'lul', 2, 'both')
@@ -77,6 +72,10 @@ def filter_ctcf_distance_table(df: pd.DataFrame, reg: str, threshold: int,
     ...
     ValueError: location parameter should be in ['upstream', \
 'downstream', 'both']
+    >>> rdf = filter_ctcf_distance_table(cdf, 'down', 0, 'both')
+    >>> rdf[['exon_name', 'dist', 'group', 'id']]
+      exon_name  dist      group   id
+    4    DSC2_5     0  siPP_DOWN  1_5
     """
     if reg not in ['down', 'up', 'all']:
         raise ValueError(f"reg parameter should be one in: "
@@ -88,12 +87,16 @@ def filter_ctcf_distance_table(df: pd.DataFrame, reg: str, threshold: int,
         df = df.loc[df["group"] == f"siPP_{reg.upper()}", :]
     if location == "upstream":
         df = df.loc[(df["dist"] >= threshold * -1) &
-                    (df["dist"] <= 0), :]
+                    (df["dist"] < 0), :]
     elif location == "downstream":
         df = df.loc[(df["dist"] <= threshold) &
-                    (df["dist"] >= 0), :]
+                    (df["dist"] > 0), :]
     else:
-        df = df.loc[abs(df["dist"]) <= threshold, :]
+        if threshold == 0:
+            df = df.loc[abs(df["dist"]) == 0, :]
+        else:
+            df = df.loc[(abs(df["dist"]) <= threshold) &
+                        (df["dist"] != 0), :]
     return df