feat: script to download and filter coriell wgs BAMs

filter_wgs_bams.py

@@ -0,0 +1,46 @@
+import os
+import logging
+from subprocess import run
+from time import sleep
+
+FILTER_BED = "brca1_palb2_chek2_atm_genes_ensembl115.bed"
+
+logging.basicConfig(
+    format="%(asctime)s [%(levelname)s] %(message)s", level=logging.INFO
+)
+
+
+def main(ftp_paths: list[str]) -> None:
+    logging.info("Script started")
+    logging.info(f"{(n_ftps:=len(ftp_paths))} BAMs to be processed")
+    logging.info(f"FTP paths: {ftp_paths}")
+
+    for idx, ftp in enumerate(ftp_paths):
+        # download
+        logging.info(f"({idx+1}/{n_ftps}) Downloading: '{ftp}'")
+        down_cmd = f"wget -nc {ftp}"
+        run(down_cmd, shell=True, capture_output=False)
+        fname = ftp.split("/")[-1]
+        logging.info(f"Download completed...")
+
+        # filtering
+        logging.info(f"Filtering '{fname}'")
+        filter_cmd = (
+            f"samtools view --bam -h -L {FILTER_BED} {fname} "
+            f"-o {fname.split(".")[0]}.filtered.bam"
+        )
+        run(filter_cmd, shell=True, capture_output=False)
+        logging.info(f"Filtering completed...")
+
+        logging.info(f"Removing large BAM {fname}")
+        os.remove(fname)
+        logging.info(f"({idx+1}/{n_ftps}) '{fname}' completed!")
+
+
+if __name__ == "__main__":
+    ftp_paths = [
+        "ftp://ftp.sra.ebi.ac.uk/vol1/run/ERR147/ERR14793899/NA14626.bam",
+        "ftp://ftp.sra.ebi.ac.uk/vol1/run/ERR147/ERR14793893/HG00343_20221005Run.bam",
+        "ftp://ftp.sra.ebi.ac.uk/vol1/run/ERR147/ERR14793896/HG03694.bam",
+    ]
+    main(ftp_paths)