diff cpt_gbk_compare/gbk_compare.py @ 0:fc603e665d75 draft default tip

Uploaded
author cpt
date Tue, 21 Jun 2022 19:46:32 +0000
parents
children
line wrap: on
line diff
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/cpt_gbk_compare/gbk_compare.py	Tue Jun 21 19:46:32 2022 +0000
@@ -0,0 +1,245 @@
+#!/usr/bin/env python3
+"""
+Copyright 2019 Ryan Wick (rrwick@gmail.com)
+https://github.com/rrwick/Compare-annotations/
+This program is free software: you can redistribute it and/or modify it under the terms of the GNU
+General Public License as published by the Free Software Foundation, either version 3 of the
+License, or (at your option) any later version.
+This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without
+even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+General Public License for more details.
+You should have received a copy of the GNU General Public License along with this program.  If not,
+see <https://www.gnu.org/licenses/>.
+"""
+
+import argparse
+from Bio import SeqIO
+from Bio import pairwise2
+from Bio.pairwise2 import format_alignment
+import itertools
+import sys
+
+
+def addArr(arrA, arrB):
+    res = []
+    for x in range(0, min(len(arrA), len(arrB))):
+      res.append(arrA[x] + arrB[x])
+    return res
+
+def get_arguments():
+    parser = argparse.ArgumentParser(description='Compare GenBank annotations',
+                                     formatter_class=argparse.ArgumentDefaultsHelpFormatter)
+
+    parser.add_argument('annotation_1', type=str,
+                        help='First annotated genome in Genbank format')
+    parser.add_argument('annotation_2', type=str,
+                        help='Second annotated genome in Genbank format')
+
+
+    parser.add_argument('--match_identity_threshold', type=float, default=0.7,
+                        help='Two genes must have at least this identity to be considerd the same (0.0 to 1.0)')
+    parser.add_argument('--allowed_skipped_genes', type=int, default=10,
+                        help='This many missing genes are allowed when aligning the annotations')
+    parser.add_argument("--addNotes", action="store_true", help="Add Note fields")
+
+    parser.add_argument(
+        "-sumOut", type=argparse.FileType("w"), help="Summary out file"
+    )
+    args = parser.parse_args()
+    return args
+
+
+def main():
+    args = get_arguments()
+
+    # Load in the CDS features from the two assemblies.
+    old = SeqIO.parse(args.annotation_1, 'genbank')
+    new = SeqIO.parse(args.annotation_2, 'genbank')
+    old_record = next(old)
+    new_record = next(new)
+    old_features, new_features = [], []
+    for f in old_record.features:
+        if f.type == 'CDS':
+            old_features.append(f)
+    for f in new_record.features:
+        if f.type == 'CDS':
+            new_features.append(f)
+
+    args.sumOut.write('Features in First Genbank\'s assembly:\t' + str(len(old_features)) + "\n")
+    args.sumOut.write('Features in Second Genbank\'s assembly:\t' + str(len(new_features)) + "\n\n")
+
+    # Align the features to each other.
+    offsets = sorted(list(itertools.product(range(args.allowed_skipped_genes + 1),
+                                            range(args.allowed_skipped_genes + 1))),
+                     key=lambda x: x[0]+x[1])
+    old_i, new_i = 0, 0
+    exactRec = 0
+    inexactRec = [0, 0 ,0]
+    hypoRec = [0, 0, 0]
+    newCount = 0
+    oldCount = 0
+    if args.addNotes:
+      print("First Record CDS Product\tSimilarity\tSecond Record CDS Product\tPercent Identity\tLength Difference\tFirst Gbk's CDS Location\tSecond Gbk's CDS Location\tHypothetical Status\tFirst Record's Notes\tSecond Record's Notes\n")
+    else:
+      print("First Record CDS Product\tSimilarity\tSecond Record CDS Product\tPercent Identity\tLength Difference\tFirst Gbk's CDS Location\tSecond Gbk's CDS Location\tHypothetical Status\n")
+    while True:
+        if old_i >= len(old_features) and new_i >= len(new_features):
+            break
+
+        for old_offset, new_offset in offsets:
+            try:
+                old_feature = old_features[old_i + old_offset]
+            except IndexError:
+                old_feature = None
+            try:
+                new_feature = new_features[new_i + new_offset]
+            except IndexError:
+                new_feature = None
+            try:
+                match, identity, length_diff = compare_features(old_feature, new_feature, old_record, new_record, args.match_identity_threshold)
+            except TypeError:
+                break
+            if match:
+                for j in range(old_offset):
+                    print_in_old_not_new(old_features[old_i + j])
+                    oldCount += 1
+                for j in range(new_offset):
+                    print_in_new_not_old(new_features[new_i + j])
+                    newCount += 1
+                if identity == 1.0:
+                  exactRec += 1
+                res1, res2 = print_match(old_features[old_i + old_offset], new_features[new_i + new_offset], identity, length_diff, args.addNotes)
+                inexactRec = addArr(inexactRec, res1)
+                hypoRec = addArr(hypoRec, res2)
+                old_i += old_offset
+                new_i += new_offset
+                break
+        else:
+            sys.stderr.write("Exceeded allowed number of skipped genes (" + str(args.allowed_skipped_genes) + "), unable to maintain alignment and continue comparison.\n")
+            exit(2)
+
+        if old_feature is None and new_feature is None:
+            break
+
+        old_i += 1
+        new_i += 1
+
+    args.sumOut.write('Exact Match:\t' + str(exactRec) + "\n\n")
+
+    args.sumOut.write('Inexact Match:\t' + str(inexactRec[0] + inexactRec[1] + inexactRec[2]) + "\n")
+    args.sumOut.write('  Same length:\t' + str(inexactRec[0]) + "\n")
+    args.sumOut.write('  Second Gbk Seq longer:\t' + str(inexactRec[2]) + "\n")
+    args.sumOut.write('  First Gbk Seq longer:\t' + str(inexactRec[1]) + "\n\n")
+     
+    args.sumOut.write('In Second Gbk but not in first:\t' + str(newCount) + "\n")
+    args.sumOut.write('In First Gbk but not in second:\t' + str(oldCount) + "\n\n")
+
+    args.sumOut.write('Hypothetical Annotation Change:\t' + str(hypoRec[1] + hypoRec[2]) + "\n")
+    args.sumOut.write('Hypothetical:\t' + str(hypoRec[0] + hypoRec[2]) + "\n")
+    
+
+def print_match(f1, f2, identity, length_diff, outNotes):
+    #print('', flush=True)
+    line = f1.qualifiers['product'][0] + "\t"
+    matchArr = [0, 0, 0]
+    hypoArr = [0, 0, 0]
+    if identity == 1.0:
+#        print('Exact match')
+        line += 'Exact match\t' + f2.qualifiers['product'][0] + "\t100.0\tSame Length\t"
+    else:
+#        print('Inexact match (' + '%.2f' % (identity * 100.0) + '% ID, ', end='')
+        line += 'Inexact match\t' + f2.qualifiers['product'][0] + "\t%.2f\t" % (identity * 100.0)
+        if length_diff == 0:
+#            print('same length)')
+            line +="Same Length\t"
+            matchArr[0] += 1
+        elif length_diff > 0:
+#            print('old seq longer)')
+            line +="First Gbk Seq Longer\t"
+            matchArr[1] += 1
+        elif length_diff < 0:
+#            print('new seq longer)')
+            line +="Second Gbk Seq Longer\t"
+            matchArr[2] += 1
+#    print('  old: ', end='')
+#    print_feature_one_line(f1)
+    line += print_feature_one_line(f1) + "\t"
+#    print('  new: ', end='')
+#    print_feature_one_line(f2)
+    line += print_feature_one_line(f2) + "\t"
+    p1 = f1.qualifiers['product'][0].lower()
+    p2 = f2.qualifiers['product'][0].lower()
+    if 'hypothetical' in p1 and 'hypothetical' in p2:
+#        print('  still hypothetical')
+        line += "Hypothetical\t"
+        hypoArr[0] += 1
+    elif 'hypothetical' in p1 and 'hypothetical' not in p2:
+#        print('  no longer hypothetical')
+        line += "No Longer Hypothetical\t"
+        hypoArr[1] += 1
+    elif 'hypothetical' not in p1 and 'hypothetical' in p2:
+#        print('  became hypothetical')
+        line += "Became Hypothetical\t"
+        hypoArr[2] += 1
+    else:
+        line += "'Hypothetical' not in second nor first Gbk's product tag"
+    
+    if outNotes:
+      line += "\t"
+      if "note" in f1.qualifiers.keys():
+        for x in f1.qualifiers["note"]:
+          line += x
+        line += "\t"
+      else:
+        line += "N/A\t"
+      if "note" in f2.qualifiers.keys():
+        for x in f2.qualifiers["note"]:
+          line += x
+      else:
+        line += "N/A"
+
+    print(line)
+    return matchArr, hypoArr
+
+def print_in_old_not_new(f): # rename file outputs
+    line = f.qualifiers['product'][0] + "\tIn First Gbk but not Second\tN/A\t0.00\t" + str(f.location.end - f.location.start) + "\t" + print_feature_one_line(f) + "\tN/A\tN/A"
+#    print('')
+#    print('In old but not in new:')
+#    print('  ', end='')
+#    print_feature_one_line(f)
+    print(line)
+
+
+def print_in_new_not_old(f): # rename file outputs
+    line = "N/A\tIn Second Gbk but not First\t" + f.qualifiers['product'][0] + "\t0.00\t" + str(f.location.end - f.location.start) + "\tN/A\t" + print_feature_one_line(f) + "\tN/A"
+    #print('')
+    #print('In new but not in old:')
+    #print('  ', end='')
+    #print_feature_one_line(f)
+    print(line)
+
+
+def print_feature_one_line(f):
+    #f_str = f.qualifiers['product'][0]
+    f_str = ""
+    strand = '+' if f.location.strand == 1 else '-'
+    f_str += '(' + str(f.location.start) + '-' + str(f.location.end) + ' ' + strand + ', '
+    f_str += str(f.location.end - f.location.start) + ' bp)'
+    return(f_str)
+
+
+def compare_features(f1, f2, r1, r2, match_identity_threshold):
+    if f1 is None or f2 is None:
+        return False
+
+    s1 = f1.extract(r1).seq
+    s2 = f2.extract(r2).seq
+    score = pairwise2.align.globalms(s1, s2, 1, 0, 0, 0, score_only=True)
+    identity = score / max(len(s1), len(s2))
+    match = identity >= match_identity_threshold
+    length_diff = len(s1) - len(s2)
+    return match, identity, length_diff
+
+
+if __name__ == '__main__':
+    main()