gzip in place

run.py

@@ -3,6 +3,7 @@ import multiprocessing
 import os
 import sys
 import shlex
+import gzip
 
 curl = """curl -X POST -H "Content-Type: application/x-www-form-urlencoded" -d 'result=sample&query=secondary_sample_accession%3D%22XXXXXX%22&format=tsv' "https://www.ebi.ac.uk/ena/portal/api/search\""""
 xml = "https://www.ebi.ac.uk/ena/browser/api/xml/"
@@ -46,24 +47,36 @@ def run_command(command):
     return result.stdout.decode('ascii')
 
 def conversion(identifier):
-    if os.path.isfile("./genome/" + identifier + "/" + identifier + ".conversion.ttl"): return
+    output_file = "./genome/" + identifier + "/" + identifier + ".conversion.ttl"
+    if os.path.isfile(output_file + ".gz"): return
     # "-tool", "conversion", "-f", fasta.getAbsolutePath(), "-gff2rdf", "-i", gff.getAbsolutePath(), "-o", output.getAbsolutePath(), "-id", "MGYG-HGUT-00001", "-codon", "11", "-topology", "linear"
-    command = "java -jar ./MGnifyParser.jar -tool conversion -f ./genome/" + identifier + "/" + identifier + ".fna -gff2rdf -i ./genome/" + identifier + "/" + identifier + ".gff -id " + identifier + " -codon 11 -topology linear -o ./genome/" + identifier + "/" + identifier + ".conversion.ttl"
+    command = "java -jar ./MGnifyParser.jar -tool conversion -f ./genome/" + identifier + "/" + identifier + ".fna -gff2rdf -i ./genome/" + identifier + "/" + identifier + ".gff -id " + identifier + " -codon 11 -topology linear -o " + output_file
     print(command)
     os.system(command)
+    gzip(output_file)
 
 def interproscan(identifier):
-    if os.path.isfile("./genome/" + identifier + "/" + identifier + ".interproscan.ttl"): return
-    command = "java -jar ./MGnifyParser.jar -tool interpro -i ./genome/" + identifier + "/" + identifier + ".conversion.ttl -o ./genome/" + identifier + "/" + identifier + ".interproscan.ttl -tsv ./genome/" +identifier +"/" + identifier + "_InterProScan.tsv -version InterProScan-v5_35-74_0"
+    output_file = "./genome/" + identifier + "/" + identifier + ".interproscan.ttl"
+    if os.path.isfile(output_file + ".gz"): return
+    command = "java -jar ./MGnifyParser.jar -tool interpro -i ./genome/" + identifier + "/" + identifier + ".conversion.ttl -o " + output_file + " -tsv ./genome/" +identifier +"/" + identifier + "_InterProScan.tsv -version InterProScan-v5_35-74_0"
     print(command)
     os.system(command)
+    gzip(output_file)
 
 def eggnog(identifier):
-    if os.path.isfile("./genome/" + identifier + "/" + identifier + ".interproscan.eggnog.ttl"): return
-    command = "java -jar ./MGnifyParser.jar -tool eggnog -i ./genome/" + identifier + "/" + identifier + ".interproscan.ttl -o ./genome/" + identifier + "/" + identifier + ".interproscan.eggnog.ttl -tsv ./genome/" +identifier +"/" + identifier + "_eggNOG.tsv -v v2-db5.0"
+    output_file = "./genome/" + identifier + "/" + identifier + ".interproscan.eggnog.ttl"
+    if os.path.isfile(output_file + ".gz"): return
+    command = "java -jar ./MGnifyParser.jar -tool eggnog -i ./genome/" + identifier + "/" + identifier + ".interproscan.ttl -o " + output_file + " -tsv ./genome/" +identifier +"/" + identifier + "_eggNOG.tsv -v v2-db5.0"
     print(command)
     os.system(command)
+    gzip(output_file)
 
+def gzip(file):
+    f_in = open(file)
+    f_out = gzip.open(file + '.gz', 'wb')
+    f_out.writelines(f_in)
+    f_out.close()
+    f_in.close()
 
 def conversion_stage(identifier):
     conversion(identifier)