src/bed_handler/select_regulated_near_ctcf_exons.py: modification of...

src/bed_handler/select_regulated_near_ctcf_exons.py: modification of create_bed_ctcf_exon function to also create a bed file for genes

src/bed_handler/select_regulated_near_ctcf_exons.py

@@ -10,6 +10,7 @@ from .config import TestConfig, BedConfig
 from .get_gene_regulated_by_ddx import load_sipp_vs_ctcf, format_exon_bed
 import pandas as pd
 from doctest import testmod
+from .filter_gene import filter_bed
 
 
 def filter_ctcf_distance_table(df: pd.DataFrame, reg: str, threshold: int,
@@ -113,9 +114,16 @@ def create_bed_ctcf_exon(reg: str, threshold: int,
                            format_exon_bed(BedConfig.exon_bed,
                                            BedConfig.gene_bed))
     df = filter_ctcf_distance_table(df, reg, threshold, location)
-    df.to_csv(BedConfig.bed.output / f"CTCF_{threshold}_{location}_"
-                                     f"ddx_{reg}_exon.bed", sep="\t",
-              index=False)
+    list_exons = df['id'].to_list()
+    list_genes = [int(exon.split('_')[0]) for exon in list_exons]
+    df_exon = filter_bed(BedConfig.exon_bed, list_exons)
+    df_gene = filter_bed(BedConfig.gene_bed, list_genes)
+    df_exon.to_csv(BedConfig.bed.output /
+                   f"CTCF_{threshold}_{location}_ddx_{reg}_exon.bed", sep="\t",
+                   index=False)
+    df_gene.to_csv(BedConfig.bed.output /
+                   f"CTCF_{threshold}_{location}_ddx_{reg}_gene.bed", sep="\t",
+                   index=False)
 
 
 if __name__ == "__main__":