update computation of the number of processes for each part of the algo (read/write/process)

microsplit/auxiliary.py

@@ -17,9 +17,10 @@ def partitionning(num_processes):
     """
     Partition the number of threads for writing and fragmenting.
     """
-    write_processes = num_processes // 4
-    compute_processes = num_processes - (write_processes * 2)
-    return write_processes, compute_processes
+    read_processes = max(int(num_processes / 3), 2)
+    write_processes = read_processes
+    compute_processes = max(num_processes - (write_processes * 2), 1)
+    return read_processes, write_processes, compute_processes
 
 def check_data(data):
     for element in data[0]:

microsplit/split.py

@@ -189,7 +189,7 @@ def cut(args, logging):
 
     input_queue = Queue()
     output_queue = Queue()
-    write_processes, compute_processes = partitionning(num_threads)
+    read_processes, write_processes, compute_processes = partitionning(num_threads)
     # communicate(write_processes=write_processes, compute_processes=compute_processes)
 
     manager = ProcessManager()
@@ -199,13 +199,13 @@ def cut(args, logging):
 
     try:
         # Start worker processes
-        manager.start_worker(target=read_bam_pair, args=(bam_for_file, bam_rev_file, input_queue, compute_processes))
+        manager.start_worker(target=read_bam_pair, args=(bam_for_file, bam_rev_file, input_queue, read_processes))
         # Process for processing items
         [
             manager.start_worker(target=process_items, args=(input_queue, output_queue, seed_size, len_add)) for _ in range(compute_processes)
         ]
         # Process for writing pairs
-        manager.start_worker(target=write_fastq_pair, args=(output_queue, output_forward, output_reverse, compute_processes))
+        manager.start_worker(target=write_fastq_pair, args=(output_queue, output_forward, output_reverse, write_processes))
         # Monitor processes
         while manager.running():
             if not manager.check_processes():