--- a/mature_mir_gff_translation.py	Mon Apr 23 13:21:16 2018 -0400
+++ b/mature_mir_gff_translation.py	Wed Apr 25 12:48:27 2018 -0400
@@ -13,105 +13,90 @@
     return args
 
 
-def get_gff_header(gff_input_file):
-    string_list = []
-    for line in open(gff_input_file, "r"):
-        if line[0] == '#':
-            string_list.append(line)
-    string_list.append('# generated by mature_mir_gff_translation.py %s\n#\n' %
-                       str(datetime.now()))
-    return ''.join(string_list)
+def convert_and_print_gff(gff_input_file, output):
 
+    def get_gff_header(gff_input_file):
+        string_list = []
+        for line in open(gff_input_file, "r"):
+            if line[0] == '#':
+                string_list.append(line)
+        string_list.append('# generated by mature_mir_gff_translation.py \
+                            %s\n#\n' % str(datetime.now()))
+        return ''.join(string_list)
 
-def load_gff_in_dict(gff_input_file):
-    '''
-    Reads the gff3 file and return a dictionary of dictionaries
-    with keys equal to standard gff3 fields (9)
-    Note that the key of the primary dictionary is the ID
-    '''
     gff_dict = {}
+    # see https://github.com/ARTbio/tools-artbio/issues/246
+    # currently fixed by perl pretreatment or the gff3 file
     for line in open(gff_input_file, "r"):
         if line[0] == "#":
             continue
         gff_fields = line[:-1].split("\t")
         ID = gff_fields[8].split("ID=")[1].split(";")[0]
-        gff_dict[ID] = {}
-        gff_dict[ID]["seqid"] = gff_fields[0]
-        gff_dict[ID]["source"] = gff_fields[1]
-        gff_dict[ID]["type"] = gff_fields[2]
-        gff_dict[ID]["start"] = gff_fields[3]
-        gff_dict[ID]["end"] = gff_fields[4]
-        gff_dict[ID]["score"] = gff_fields[5]
-        gff_dict[ID]["strand"] = gff_fields[6]
-        gff_dict[ID]["phase"] = gff_fields[7]
-        gff_dict[ID]["attributes"] = gff_fields[8]
-        if "erives_from" in gff_dict[ID]["attributes"]:
-            parent_primary_transcript = gff_dict[ID]["attributes"].split(
-                "erives_from=")[1]
-            parent_primary_transcript = gff_dict[parent_primary_transcript][
-                                            "attributes"].split("Name=")[1]
-            gff_dict[ID]["attributes"] = "%s;Parent_mir_Name=%s" % (
-                gff_dict[ID]["attributes"], parent_primary_transcript)
-    return gff_dict
-
-
-def genome_to_mir_gff(gff_dict, output, header):
-    '''
-    Converts seqid field from chromosome to item Name
-    Then converts coordinates relative to "miRNA_primary_transcript"
-    Note that GFF files are 1-based coordinates
-    '''
-    for key in gff_dict:
-        name = gff_dict[key]["attributes"].split("Name=")[1].split(";")[0]
-        gff_dict[key]["seqid"] = name
-        if "erives_from=" in gff_dict[key]["attributes"]:
-            parent_ID = gff_dict[key]["attributes"].split(
-                                        "erives_from=")[1].split(";")[0]
-            gff_dict[key]["start"] = str(int(gff_dict[key]["start"])-int(
-                                       gff_dict[parent_ID]["start"])+1)
-            gff_dict[key]["end"] = str(int(gff_dict[key]["end"])-int(
-                                     gff_dict[parent_ID]["start"])+1)
-    hairpins = {}
-    matures = {}
-    # treats miRNA_primary_transcript coordinates
-    # in a second loop to avoid errors in conversion
-    for key in gff_dict:
-        if gff_dict[key]["type"] == "miRNA_primary_transcript":
-            gff_dict[key]["end"] = str(int(gff_dict[key]["end"])-int(
-                                       gff_dict[key]["start"]) + 1)
-            gff_dict[key]["start"] = '1'
-            # now, do a dict[ID]=Name but only for miRNA_primary_transcript
-            hairpins[key] = gff_dict[key]["attributes"].split(
-                                                        "Name=")[1].split(
-                                                        ";")[0]
+        if gff_fields[2] == "miRNA_primary_transcript":
+            gff_dict[ID] = {}
+            gff_dict[ID]["premir_name"] = gff_fields[8].split(
+                "Name=")[1].split(";")[0]
+            gff_dict[ID]["primary"] = line[:-1]
+            gff_dict[ID]["miRNAs"] = []
+        elif gff_fields[2] == "miRNA":
+            parent_ID = gff_fields[8].split("erives_from=")[1]
+            gff_dict[parent_ID]["miRNAs"].append(line[:-1])
+    # Now reorganise features and recalculate coordinates of premirs and mirs
+    gff_list = []
+    for ID in sorted(gff_dict, key=lambda x: (gff_dict[x]['premir_name'])):
+        # delete premir and their mir with ID containing "_"
+        if "_" in ID:
+            del gff_dict[ID]
         else:
-            matures[key] = gff_dict[key]["attributes"].split(
-                                                        "Name=")[1].split(
-                                                         ";")[0]
+            primary_fields = gff_dict[ID]["primary"].split('\t')
+            seqid = primary_fields[8].split("Name=")[1].split(";")[0]
+            source = primary_fields[1]
+            type = primary_fields[2]
+            start = primary_fields[3]
+            newstart = "1"
+            end = primary_fields[4]
+            newend = str(int(end)-int(start)+1)
+            score = primary_fields[5]
+            strand = primary_fields[6]
+            phase = primary_fields[7]
+            attributes = primary_fields[8]
+            gff_list.append('%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s' % (seqid,
+                            source, type, newstart, newend, score, strand,
+                            phase, attributes))
+            # ensure their is only 2 child miRNAs at best
+            if len(gff_dict[ID]["miRNAs"]) > 2:
+                gff_dict[ID]["miRNAs"] = gff_dict[ID]["miRNAs"][:2]
+            # sort child miRNAs 5p first 3p second
+            if gff_dict[ID]["miRNAs"][0].find('5p') == -1:
+                gff_dict[ID]["miRNAs"] = gff_dict[ID]["miRNAs"][::-1]
+            for mir in gff_dict[ID]["miRNAs"]:
+                mir_fields = mir.split('\t')
+                mir_seqid = mir_fields[8].split("Name=")[1].split(";")[0]
+                mir_source = mir_fields[1]
+                mir_type = mir_fields[2]
+                mir_start = mir_fields[3]
+                mir_end = mir_fields[4]
+                new_mir_start = str(int(mir_start)-int(start)+1)
+                new_mir_end = str(int(mir_end)-int(start)+1)
+                mir_score = mir_fields[5]
+                mir_strand = mir_fields[6]
+                mir_phase = mir_fields[7]
+                mir_attributes = mir_fields[8]
+                mir_sfx = ";Parent_mir_Name=%s" % gff_dict[ID]["premir_name"]
+                gff_list.append(
+                                '%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s\t%s%s' % (
+                                    mir_seqid, mir_source, mir_type,
+                                    new_mir_start, new_mir_end, mir_score,
+                                    mir_strand, mir_phase, mir_attributes,
+                                    mir_sfx))
     with open(output, "w") as output:
-        output.write(header)
-        for ID in sorted(hairpins, key=hairpins.get):
-            output.write("\t".join([gff_dict[ID]["seqid"],
-                         gff_dict[ID]["source"], gff_dict[ID]["type"],
-                         gff_dict[ID]["start"], gff_dict[ID]["end"],
-                         gff_dict[ID]["score"], gff_dict[ID]["strand"],
-                         gff_dict[ID]["phase"], gff_dict[ID]["attributes"]]))
-            output.write("\n")
-            for id in sorted(matures, key=matures.get, reverse=True):
-                if ID in gff_dict[id]["attributes"]:
-                    output.write("\t".join([gff_dict[id]["seqid"],
-                                 gff_dict[id]["source"], gff_dict[id]["type"],
-                                 gff_dict[id]["start"], gff_dict[id]["end"],
-                                 gff_dict[id]["score"],
-                                 gff_dict[id]["strand"],
-                                 gff_dict[id]["phase"],
-                                 gff_dict[id]["attributes"]]))
-                    output.write("\n")
+        output.write('%s' % get_gff_header(gff_input_file))
+        output.write('\n'.join(gff_list))
+        output.write('\n')
 
 
 def main(infile, outfile):
-    gff_dict = load_gff_in_dict(infile)
-    genome_to_mir_gff(gff_dict, outfile, get_gff_header(infile))
+    convert_and_print_gff(infile, outfile)
 
 
 if __name__ == "__main__":