Mercurial > repos > galaxyp > peptide_genomic_coordinate

annotate peptide_genomic_coordinate.py @ 1:cb0378d2d487 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
author galaxyp
date Sun, 14 Mar 2021 03:01:11 +0000
parents 5f49ffce52cb
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
1 #!/usr/bin/env python
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
2 #
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
3 # Author: Praveen Kumar
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
4 # University of Minnesota
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
5 #
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
6 # Get peptide's genomic coordinate from the protein's genomic mapping sqlite file
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
7 # (which is derived from the https://toolshed.g2.bx.psu.edu/view/galaxyp/translate_bed/038ecf54cbec)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
8 #
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
9 # python peptideGenomicCoordinate.py <peptide_list> <mz_to_sqlite DB> <genomic mapping file DB> <output.bed>
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
10 #
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
11 import argparse
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
12 import sqlite3
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
13 import sys
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
14
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
15
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
16 pep_stmt = """\
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
17 SELECT dBSequence_ref, start, end, peptide_ref \
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
18 FROM peptide_evidence e JOIN peptides p on e.peptide_ref = p.id \
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
19 WHERE isDecoy = 'false' AND p.sequence = ?\
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
20 """
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
21
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
22 map_stmt = """
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
23 SELECT name, chrom, start, end, strand, cds_start, cds_end \
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
24 FROM feature_cds_map \
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
25 WHERE name = ? \
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
26 AND cds_end >= ? AND cds_start <= ? \
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
27 ORDER BY cds_start\
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
28 """
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
29
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
30
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
31 def main():
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
32 parser = argparse.ArgumentParser(description='BED file for peptides')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
33 parser.add_argument('peptides_file',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
34 metavar='peptides.tabular',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
35 type=argparse.FileType('r'),
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
36 help='List of peptides, one per line')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
37 parser.add_argument('mz_to_sqlite',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
38 metavar='mz.sqlite',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
39 help='mz_to_sqlite sqlite database')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
40 parser.add_argument('genome_mapping',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
41 metavar='genome_mapping.sqlite',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
42 help='genome_mapping sqlite database')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
43 parser.add_argument('bed_file',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
44 metavar='peptides.bed',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
45 type=argparse.FileType('w'),
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
46 help='BED file of peptide genomic locations')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
47 parser.add_argument('-a', '--accession',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
48 action='store_true',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
49 help='Append the accession to the peptide for BED name')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
50 parser.add_argument('-d', '--debug',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
51 action='store_true',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
52 help='Debug')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
53 args = parser.parse_args()
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
54
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
55 pconn = sqlite3.connect(args.mz_to_sqlite)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
56 pc = pconn.cursor()
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
57 mconn = sqlite3.connect(args.genome_mapping)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
58 mc = mconn.cursor()
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
59 outfh = args.bed_file
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
60 pepfile = args.peptides_file
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
61 for seq in pepfile.readlines():
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
62 seq = seq.strip()
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
63 pc.execute(pep_stmt, (seq,))
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
64 pep_refs = pc.fetchall()
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
65 for pep_ref in pep_refs:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
66 (acc, pep_start, pep_end, pep_seq) = pep_ref
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
67 cds_start = (pep_start - 1) * 3
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
68 cds_end = pep_end * 3
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
69 if args.debug:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
70 print('%s\t%s\t%s\t%d\t%d' % (acc, pep_start, pep_end, cds_start, cds_end), file=sys.stdout)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
71 mc.execute(map_stmt, (acc, cds_start, cds_end))
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
72 exons = mc.fetchall()
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
73 if args.debug:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
74 print('\n'.join([str(e) for e in exons]), file=sys.stdout)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
75 if exons:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
76 chrom = exons[0][1]
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
77 strand = exons[0][4]
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
78 if strand == '+':
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
79 start = exons[0][2] + cds_start
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
80 end = exons[-1][2] + cds_end - exons[-1][5]
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
81 blk_start = []
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
82 blk_size = []
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
83 for exon in exons:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
84 offset = cds_start if cds_start > exon[5] else 0
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
85 bstart = exon[2] + offset
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
86 bsize = min(cds_end, exon[6]) - max(cds_start, exon[5])
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
87 if args.debug:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
88 print('bstart %d\tbsize %d\t %d' % (bstart, bsize, offset), file=sys.stdout)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
89 blk_start.append(bstart - start)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
90 blk_size.append(bsize)
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
91 else:
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
92 start = exons[-1][2] + exons[-1][6] - cds_end
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
93 end = exons[0][3] - cds_start + exons[0][5]
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
94 blk_start = []
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
95 blk_size = []
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
96 for exon in reversed(exons):
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
97 bstart = exon[2] + exon[6] - min(exon[6], cds_end)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
98 bsize = min(cds_end, exon[6]) - max(cds_start, exon[5])
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
99 # bend = exon[3] - (exon[5] - max(exon[5], cds_start))
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
100 bend = exon[3] - min(cds_start - exon[5], cds_start)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
101 bend = exon[3] - bsize
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
102 if args.debug:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
103 print('bstart %d\tbsize %d\tbend %d' % (bstart, bsize, bend), file=sys.stdout)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
104 blk_start.append(bstart - start)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
105 blk_size.append(bsize)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
106 bed_line = [str(chrom), str(start), str(end),
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
107 '_'.join([seq, acc]) if args.accession else seq,
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
108 '255', strand,
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
109 str(start), str(end),
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
110 '0,0,0',
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
111 str(len(blk_start)),
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
112 ','.join([str(b) for b in blk_size]),
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
113 ','.join([str(b) for b in blk_start])]
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
114 if args.debug:
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
115 print('\t'.join(bed_line), file=sys.stdout)
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
116 outfh.write('\t'.join(bed_line) + '\n')
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
117 pconn.close()
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
118 mconn.close()
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
119 outfh.close()
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
120 pepfile.close()
1
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
121
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
122
cb0378d2d487 "planemo upload commit 43b42fdeef93a498e893fe13f99a076af294e603"
galaxyp
parents: 0
diff changeset
123 if __name__ == '__main__':
0
5f49ffce52cb planemo upload commit be7e9677908b7864ef0b965a1e219a1840eeb2ec
galaxyp
parents:
diff changeset
124 main()