domain annotation is now sorted by sequence-id and sequence-start for simple architecture output

scripts/simple_arch.py

@@ -147,25 +147,24 @@ def main():
             taxid2name[taxid] = name
 
     with open(finalPredictionsFile,'w') as outFile:
-        for query in objReturn:
-            for i in range(0,len(objReturn[query])):
-                if not objReturn[query][i] is None:
-                    data     = objReturn[query][i]
-                    evalue   = data[I_EVALUE]
-                    bitscore = data[SCORE]
-                    modid    = data[MODEL_NAME]
-                    modbeg   = data[MODEL_START]
-                    modend   = data[MODEL_END]
-                    modlen   = data[MODEL_LEN]
-                    domid    = data[TARGET_ACC]
-                    seqid    = data[QUERY_NAME]
-                    seqbeg   = data[SEQ_START]
-                    seqend   = data[SEQ_END]
-                    seqlen   = data[SEQ_LEN]
-                    accuracy = data[ACCURACY]
-                    modtaxid = data[MODEL_TAXID]
-                    modtaxname = taxid2name.get(modtaxid,'unavailable')
-                    outFile.write(f'{seqid}\t{seqbeg}\t{seqend}\t{seqlen}\t{domid}\t{modid}\t{modbeg}\t{modend}\t{modlen}\t{evalue}\t{bitscore}\t{accuracy}\t{modtaxname}\n')
+        for query in sorted(objReturn.keys()):
+            sortedHits = sorted( (data for data in objReturn[query] if data is not None), key=lambda d:d[SEQ_START] )
+            for data in sortedHits:
+                evalue   = data[I_EVALUE]
+                bitscore = data[SCORE]
+                modid    = data[MODEL_NAME]
+                modbeg   = data[MODEL_START]
+                modend   = data[MODEL_END]
+                modlen   = data[MODEL_LEN]
+                domid    = data[TARGET_ACC]
+                seqid    = data[QUERY_NAME]
+                seqbeg   = data[SEQ_START]
+                seqend   = data[SEQ_END]
+                seqlen   = data[SEQ_LEN]
+                accuracy = data[ACCURACY]
+                modtaxid = data[MODEL_TAXID]
+                modtaxname = taxid2name.get(modtaxid,'unavailable')
+                outFile.write(f'{seqid}\t{seqbeg}\t{seqend}\t{seqlen}\t{domid}\t{modid}\t{modbeg}\t{modend}\t{modlen}\t{evalue}\t{bitscore}\t{accuracy}\t{modtaxname}\n')
     # END OF ARI CODE #
     runtime = time.time()-start_time
     eprint(f'[{script_name}] runtime: {runtime:.2f}')