parallel added

run.py

+from joblib import Parallel, delayed
+import multiprocessing
 import os
 import sys
 import shlex
@@ -62,12 +64,22 @@ def eggnog(identifier):
     print(command)
     os.system(command)
 
+
+def conversion_stage(identifier):
+    conversion(identifier)
+    interproscan(identifier)
+    eggnog(identifier)
+
 if __name__ == "__main__":
+    identifiers = set()
     for line in open("identifiers.tsv"):
+        if "MGYG" not in line: continue
         identifier = line.strip().split("\t")[1]
         accession =  line.strip().split("\t")[2]
         if (identifier.startswith("MGYG")):
             retrieval(identifier, accession)
-            conversion(identifier)
-            interproscan(identifier)
-            eggnog(identifier)
+            identifiers.add(identifier)
+
+    # Number of cores
+    num_cores = 3 # multiprocessing.cpu_count()
+    results = Parallel(n_jobs=num_cores)(delayed(conversion_stage)(identifier) for identifier in identifiers)