Merge pull request #36 from genomewalker:fix-sorted-location

Updated README

bam_filter/filter.py

@@ -93,12 +93,12 @@ def filter_references(args):
         mode="filter",
     )
 
-    sorted_bam = args.bam.replace(".bam", ".bf-sorted.bam")
     bam = check_bam_file(
         bam=args.bam,
         threads=4,
         reference_lengths=args.reference_lengths,
         sort_memory=args.sort_memory,
+        sorted_bam=out_files["sorted_bam"],
     )
     if bam is None:
         logging.warning("No reference sequences with alignments found in the BAM file")
@@ -264,11 +264,4 @@ def filter_references(args):
     if args.low_memory:
         os.remove(out_files["bam_tmp_sorted"])
     # check if sorted BAM file exists, if yes remove it
-    if os.path.exists(sorted_bam):
-        os.remove(sorted_bam)
-    # check if sorted BAM index file exists, if yes remove it
-    if os.path.exists(sorted_bam + ".bai"):
-        os.remove(sorted_bam + ".bai")
-    elif os.path.exists(sorted_bam + ".csi"):
-        os.remove(sorted_bam + ".csi")
     logging.info("ALL DONE.")

bam_filter/reassign.py

@@ -966,12 +966,12 @@ def reassign(args):
         mode="reassign",
         bam_reassigned=args.bam_reassigned,
     )
-    sorted_bam = bam.replace(".bam", ".bf-sorted.bam")
     bam = check_bam_file(
         bam=args.bam,
         threads=args.threads,
         reference_lengths=args.reference_lengths,
         sort_memory=args.sort_memory,
+        sorted_bam=out_files["sorted_bam"],
     )
     if bam is None:
         logging.warning("No reference sequences with alignments found in the BAM file")
@@ -1009,10 +1009,5 @@ def reassign(args):
         disable_sort=args.disable_sort,
         tmp_dir=tmp_dir,
     )
-    if os.path.exists(sorted_bam):
-        os.remove(sorted_bam)
-    if os.path.exists(sorted_bam + ".bai"):
-        os.remove(sorted_bam + ".bai")
-    elif os.path.exists(sorted_bam + ".csi"):
-        os.remove(sorted_bam + ".csi")
+
     log.info("Done!")

bam_filter/sam_utils.py

@@ -788,11 +788,12 @@ def check_bam_file(
     threads=1,
     reference_lengths=None,
     sort_memory="1G",
+    sorted_bam=None,
 ):
     logging.info("Checking BAM file status")
     save = pysam.set_verbosity(0)
 
-    def process_bam(bam, s_threads):
+    def evaluate_bam(bam, s_threads, sorted_bam=None):
         with pysam.AlignmentFile(bam, "rb", threads=s_threads) as samfile:
             references = samfile.references
             log.info(f"::: Found {samfile.nreferences:,} reference sequences")
@@ -816,7 +817,8 @@ def process_bam(bam, s_threads):
 
             if samfile.header["HD"]["SO"] != "coordinate":
                 log.info("::: BAM file is not sorted by coordinates, sorting it...")
-                sorted_bam = bam.replace(".bam", ".bf-sorted.bam")
+                if sorted_bam is None:
+                    sorted_bam = bam.replace(".bam", ".bf-sorted.bam")
                 pysam.sort(
                     "-@", str(s_threads), "-m", str(sort_memory), "-o", sorted_bam, bam
                 )
@@ -833,11 +835,11 @@ def process_bam(bam, s_threads):
 
     try:
         s_threads = min(threads, 4)
-        bam, reopened = process_bam(bam, s_threads)
+        bam, reopened = evaluate_bam(bam, s_threads, sorted_bam)
 
         # If the BAM file was sorted and reopened, check it again
         if reopened:
-            bam, _ = process_bam(bam, s_threads)
+            bam, _ = evaluate_bam(bam, s_threads, sorted_bam)
 
         pysam.set_verbosity(save)
         return bam

bam_filter/utils.py

@@ -1520,6 +1520,7 @@ def create_output_files(
         log.error("Mode not recognized")
         exit(1)
     out_files["tmp_dir"] = tmp_dir
+    out_files["sorted_bam"] = f"{tmp_dir}/{prefix}.bf-sorted.bam"
     return out_files
 
     # out_files = {