Added d-tartrate qc message for Paratyphi B, Paratyphi B var. Java an…

…d I 1,4,[5],12:i:-

README.rst

@@ -221,9 +221,9 @@ If you run ``sistr -h``, you should see the following usage info:
 	                        Number of parallel threads to run sistr_cmd analysis.
 	  -l LIST_OF_SEROVARS, --list-of-serovars LIST_OF_SEROVARS
                         	A path to a single column text file containing list of
-                        	serovar(s) to check serovar prediction against. Report
-                        	predicted serovar is Y (present) and N (absent) in the
-                        	list
+                        	serovar(s) to check serovar prediction against. Result
+                        	reported in the "predicted_serovar_in_list"
+                        	field as Y (present) or N (absent) value.
 	  -v, --verbose         Logging verbosity level (-v == show warnings; -vvv ==
 	                        show debug info)
 	  -V, --version         show program's version number and exit

sistr/sistr_cmd.py

@@ -95,7 +95,7 @@ def init_parser():
                         help='Number of parallel threads to run sistr_cmd analysis.')
     parser.add_argument('-l', '--list-of-serovars', nargs='?',
                         required=False, const=os.path.join(os.path.dirname(os.path.abspath(__file__)), "data/serovar-list.txt"),
-                        help='A path to a single column text file containing list of serovar(s) to check SISTR serovar prediction against. Report predicted serovar in "serovar_in_serovar_list" field as Y (present) or N (absent).')
+                        help='A path to a single column text file containing list of serovars to check SISTR serovar prediction against. Result reported in the "predicted_serovar_in_list" field as Y (present) or N (absent) value.')
     parser.add_argument('-v',
                         '--verbose',
                         action='count',
@@ -255,7 +255,7 @@ def sistr_predict(input_fasta, genome_name, tmp_dir, keep_tmp, args):
         if os.path.exists(args.list_of_serovars):
             with open(args.list_of_serovars) as fp:
                 serovars_selected_list = [l.rstrip() for l in fp.readlines()]
-            logging.info(f"Selected list of serovars at {args.list_of_serovars} with {len(serovars_selected_list)} serovars to check SISTR serovar prediction against. Result reported in the 'serovar_in_serovar_list' field")
+            logging.info(f"Using the selected list of serovars {args.list_of_serovars} with {len(serovars_selected_list)} serovars to check overall SISTR serovar prediction against. Result will be reported in the 'predicted_serovar_in_list' field")
         else:
             logging.warning(f"File {args.list_of_serovars} does not exist in path specified. Would not perform SISTR serovar check against the list of serovars ...")      
 
@@ -297,10 +297,11 @@ def sistr_predict(input_fasta, genome_name, tmp_dir, keep_tmp, args):
         infer_o_antigen(prediction)
         # if list of reportable serovars is provided to check prediction serovar against
         if serovars_selected_list:
-            prediction.serovar_in_serovar_list = "N"
-            for selected_serovar in serovars_selected_list:
-                if selected_serovar == prediction.serovar:
-                    prediction.serovar_in_serovar_list = "Y"
+            prediction.predicted_serovar_in_list = "N"
+            for serovar in serovars_selected_list:
+                if serovar in prediction.serovar: #try to match list serovar to the predicted serovar(s) 
+                    prediction.predicted_serovar_in_list = "Y"
+                    logging.info(f"Found {serovar} serovar from {args.list_of_serovars} in SISTR predicted serovar(s) ({prediction.serovar})")
                     break
 
         logging.info('%s | Antigen gene BLAST serovar prediction: "%s" serogroup=%s %s:%s:%s',

sistr/src/qc/__init__.py

@@ -53,5 +53,9 @@ def qc(fasta_path, cgmlst_results, prediction):
         qc_status = 'FAIL'
         qc_msgs.append('FAIL: Wzx/Wzy genes missing. Cannot determine O-antigen group/serogroup. Cannot accurately predict serovar from antigen genes.')
 
+    if prediction.serovar in 'Paratyphi B|Paratyphi B var. Java':
+        qc_msgs.append("Perform d-tartrate test (dT) to differentiate between Paratyphi B and Paratyphi B var. Java. The dT+ result is indicative of variant Java.")
+    if prediction.serovar in 'I 1,4,[5],12:i:-':
+        qc_msgs.append(f"Perform d-tartrate test (dT) as both dT+ and dT- {prediction.serovar} subtypes exist.")
     qc_msgs.sort()
     return qc_status, qc_msgs