Generate job queue correctly for bam with unsorted contig lengths

singlecellmultiomics/universalBamTagger/bamtagmultiome.py

@@ -333,18 +333,15 @@ def tag_multiome_multi_processing(
         small_contig_threshold = 100000
 
         job_gen =  [[('*',None,None,None,None),],]
-        current = []
+        small_contigs_jobs = []
         for contig,contig_len in get_contigs_with_reads(input_bam_path, True):
             if contig_len<small_contig_threshold:
-                current.append( (contig,None,None,None,None) )
+                small_contigs_jobs.append( (contig,None,None,None,None) )
             else:
-                if len(current)>1:
-                    job_gen.append(current)
-                    current=[]
-                else:
-                    job_gen.append([ (contig,None,None,None,None), ])
-        if len(current)>1:
-            job_gen.append(current)
+                job_gen.append([ (contig,None,None,None,None), ])
+        if len(small_contigs_jobs)>0:
+            job_gen.append(small_contigs_jobs)
+
         #job_gen =  [[('*',None,None,None,None),],] + [ [(contig,None,None,None,None),] for contig,contig_len in get_contigs_with_reads(input_bam_path, True) if contig!='*' ]
 
 
@@ -375,6 +372,7 @@ def tag_multiome_multi_processing(
                            additional_args= additional_args,
                            max_time_per_segment=max_time_per_segment)
 
+
     # Create header bam:
     temp_header_bam_path = f'{temp_folder}/{uuid.uuid4()}_header.bam'
     with pysam.AlignmentFile(input_bam_path) as input_bam:

singlecellmultiomics/universalBamTagger/tagging.py

@@ -158,6 +158,15 @@ def timeout_check_function(iteration, mol_iter, reads ):
             'time_start': time_start}
 
 
+def task_to_alias(kwargs):
+    if kwargs.get('contig') is None:
+        return 'all'
+    contig = kwargs['contig'].replace('*', 'star')
+    if kwargs.get('start') is None:
+        return contig
+    if kwargs.get('end') is None:
+        return contig + '-' + kwargs['start']
+    return contig + '-' + kwargs['start'] + '-' + kwargs['end']
 
 def run_tagging_tasks(args: tuple):
     """ Run tagging for one or more tasks
@@ -169,11 +178,10 @@ def run_tagging_tasks(args: tuple):
 
     (alignments_path, temp_dir, timeout_time), arglist = args
 
-    target_file = f"{temp_dir}/{uuid4()}.bam"
-    while os.path.exists(target_file):
-        print(f'Collision at {target_file}')
-        target_file = f"{temp_dir}/{uuid4()}.bam"
 
+    target_file_prefix = '_'.join( [task_to_alias(kwargs)  for kwargs in arglist[:2]]) + (f'_{len(arglist)-2}more' if len(arglist) > 2 else '' )
+    target_file = f"{temp_dir}/{target_file_prefix}.bam"
+    assert not os.path.exists(target_file), f'File {target_file} already exists'
 
     timeout_tasks = []
     total_molecules = 0

singlecellmultiomics/version.py

@@ -1,2 +1,2 @@
-__version__ = '0.1.44'
+__version__ = '0.1.45'