From f9647cba1a6a29ab0e55e5b82f74e585825c6d92 Mon Sep 17 00:00:00 2001
From: Fontrodona Nicolas <nicolas.fontrodona@ens-lyon.fr>
Date: Mon, 26 Oct 2020 09:46:15 +0100
Subject: [PATCH] src/bed_handler/select_regulated_near_ctcf_exons.py:
modification of create_bed_ctcf_exon function to also create a bed file for
genes
---
.../select_regulated_near_ctcf_exons.py | 14 +++++++++++---
1 file changed, 11 insertions(+), 3 deletions(-)
diff --git a/src/bed_handler/select_regulated_near_ctcf_exons.py b/src/bed_handler/select_regulated_near_ctcf_exons.py
index 8c1bab0..c5fcf6c 100644
--- a/src/bed_handler/select_regulated_near_ctcf_exons.py
+++ b/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__":
--
GitLab