v1.0.4

Characters "|" and "@" are not allowed in query ids

aux_scripts/calculate_ai.py

@@ -40,6 +40,8 @@ def main():
     egp_taxid.add(2787823)
     egp_taxid.add(2787854)
 
+    forbidden_chars = ["|", "@"]
+
     #Setting up NCBI Taxonomy
     ncbi = NCBITaxa()
     number_of_lost_taxids = 0
@@ -64,6 +66,8 @@ def main():
                 skip += 1
             else:
                 gene = elements[0]
+                if any(char in gene for char in forbidden_chars):
+                    sys.exit("[x] Query Ids should not contain | or @ characters")
                 list_genes.append(gene)
                 hit = elements[1]
                 iden = elements[2]

avp

@@ -1,5 +1,5 @@
 #!/usr/bin/env python3
-#v1.0.3
+#v1.0.4
 
 from depot.interface import main
 

depot/evaluate.py

@@ -126,7 +126,6 @@ def main():
             con = open(con_path,'w')
 
             FNULL = open(os.devnull, 'w')
-            all_trees_path =  os.path.join(out_path,gene + ".trees")
 
             if fastml:
                 # Constaint file for FastTree

depot/interface.py

@@ -23,7 +23,7 @@
 from docopt import docopt
 
 def main():
-    args = docopt(__doc__,version='1.0.3', options_first=True)
+    args = docopt(__doc__,version='1.0.4', options_first=True)
     if args['<module>'] == 'prepare':
         import depot.prepare as prepare
         prepare.main()

depot/prepare.py

@@ -113,6 +113,8 @@ def main():
     selection = selection.replace("ahs","float(row[i_ahs])>"+str(ahs_cutoff))
     selection = selection.replace("outg_pct","float(row[i_pct])>"+str(pct_cutoff))
 
+    forbidden_chars = ["|", "@"]
+
     if trim:
         check_programs("trimal")
         trim_folder = os.path.join(output_dir,"trim")
@@ -127,7 +129,7 @@ def main():
     elif data_type == "DNA":
         dbtype = "nucl"
     else:
-        sys.exit("data_type should be either AA or DNA")
+        sys.exit("[x] data_type should be either AA or DNA")
 
     #Setting up NCBI Taxonomy
     ncbi = NCBITaxa()
@@ -185,6 +187,8 @@ def main():
             if('#' not in line):
                 L_hitqline = line.rstrip('\n').split('\t')
                 query_id = L_hitqline[0]
+                if any(char in query_id for char in forbidden_chars):
+                    sys.exit("[x] Query Ids should not contain | or @ characters")
                 if query_id in queries_info.keys(): # Queries that pass the initial selection
                     if(len(query_dict_set[query_id]) <= queries_info[query_id]["pos"]):
                         query_hit_id = L_hitqline[1]