Mercurial > repos > galaxyp > pep_pointer

annotate pep_pointer.py @ 3:a6282baa8c6f draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
author galaxyp
date Mon, 20 Jun 2022 13:59:52 +0000
parents 073a2965e3b2
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
1
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
2 #
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
3 # Author: Praveen Kumar
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
4 # Updated: April 6th, 2018 (updated to python3: May 2022)
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
5 #
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
6 #
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
7 #
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
8
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
9 import re
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
10
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
11
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
12 def main():
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
13 import sys
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
14 if len(sys.argv) == 4:
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
15 inputFile = sys.argv
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
16 infh = open(inputFile[1], "r")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
17 # infh = open("Mus_musculus.GRCm38.90.chr.gtf", "r")
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
18
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
19 gtf = {}
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
20 gtf_transcript = {}
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
21 gtf_gene = {}
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
22 for each in infh.readlines():
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
23 a = each.split("\t")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
24 if re.search("^[^#]", each):
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
25 if re.search("gene_biotype \"protein_coding\"", a[8]) and int(a[4].strip()) != int(a[3].strip()):
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
26 type = a[2].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
27 if type == "gene" or type == "exon" or type == "CDS" or type == "five_prime_utr" or type == "three_prime_utr":
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
28 chr = "chr" + a[0].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
29 strand = a[6].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
30 if strand == "+":
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
31 start = a[3].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
32 end = a[4].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
33 elif strand == "-":
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
34 if int(a[4].strip()) > int(a[3].strip()):
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
35 start = a[3].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
36 end = a[4].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
37 elif int(a[4].strip()) < int(a[3].strip()):
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
38 start = a[4].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
39 end = a[3].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
40 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
41 print("Please check the start end coordinates in the GTF file")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
42 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
43 print("Please check the strand information in the GTF file. It should be '+' or '-'.")
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
44 if strand not in gtf:
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
45 gtf[strand] = {}
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
46 if type not in gtf[strand]:
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
47 gtf[strand][type] = []
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
48 b = re.search("gene_id \"(.+?)\";", a[8].strip())
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
49 gene = b.group(1)
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
50 if type == "gene":
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
51 transcript = ""
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
52 else:
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
53 b = re.search("transcript_id \"(.+?)\";", a[8].strip())
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
54 transcript = b.group(1)
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
55 data = (chr, start, end, gene, transcript, strand, type)
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
56 gtf[strand][type].append(data)
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
57
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
58 if type == "exon":
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
59 if chr + "#" + strand in gtf_transcript:
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
60 if transcript + "#" + gene in gtf_transcript[chr + "#" + strand]:
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
61 gtf_transcript[chr + "#" + strand][transcript + "#" + gene][0].append(int(start))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
62 gtf_transcript[chr + "#" + strand][transcript + "#" + gene][1].append(int(end))
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
63 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
64 gtf_transcript[chr + "#" + strand][transcript + "#" + gene] = [[], []]
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
65 gtf_transcript[chr + "#" + strand][transcript + "#" + gene][0].append(int(start))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
66 gtf_transcript[chr + "#" + strand][transcript + "#" + gene][1].append(int(end))
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
67 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
68 gtf_transcript[chr + "#" + strand] = {}
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
69 gtf_transcript[chr + "#" + strand][transcript + "#" + gene] = [[], []]
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
70 gtf_transcript[chr + "#" + strand][transcript + "#" + gene][0].append(int(start))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
71 gtf_transcript[chr + "#" + strand][transcript + "#" + gene][1].append(int(end))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
72
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
73 if type == "gene":
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
74 if chr + "#" + strand in gtf_gene:
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
75 gtf_gene[chr + "#" + strand][0].append(int(start))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
76 gtf_gene[chr + "#" + strand][1].append(int(end))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
77 gtf_gene[chr + "#" + strand][2].append(gene)
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
78 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
79 gtf_gene[chr + "#" + strand] = [[0], [0], ["no_gene"]]
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
80 gtf_gene[chr + "#" + strand][0].append(int(start))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
81 gtf_gene[chr + "#" + strand][1].append(int(end))
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
82 gtf_gene[chr + "#" + strand][2].append(gene)
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
83
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
84 # "Starting Reading Intron . . ."
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
85
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
86 gtf["+"]["intron"] = []
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
87 gtf["-"]["intron"] = []
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
88 for chr_strand in gtf_transcript.keys():
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
89 chr = chr_strand.split("#")[0]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
90 strand = chr_strand.split("#")[1]
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
91
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
92 for transcript_gene in gtf_transcript[chr_strand].keys():
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
93 start_list = gtf_transcript[chr_strand][transcript_gene][0]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
94 end_list = gtf_transcript[chr_strand][transcript_gene][1]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
95 sorted_start_index = [i[0] for i in sorted(enumerate(start_list), key=lambda x:x[1])]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
96 sorted_end_index = [i[0] for i in sorted(enumerate(end_list), key=lambda x:x[1])]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
97 if sorted_start_index == sorted_end_index:
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
98 sorted_start = sorted(start_list)
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
99 sorted_end = [end_list[i] for i in sorted_start_index]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
100 for x in range(len(sorted_start))[1:]:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
101 intron_start = sorted_end[x - 1] + 1
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
102 intron_end = sorted_start[x] - 1
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
103 transcript = transcript_gene.split("#")[0]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
104 gene = transcript_gene.split("#")[1]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
105 data = (chr, str(intron_start), str(intron_end), gene, transcript, strand, "intron")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
106 gtf[strand]["intron"].append(data)
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
107
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
108 # "Starting Reading Intergenic . . ."
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
109
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
110 gtf["+"]["intergenic"] = []
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
111 gtf["-"]["intergenic"] = []
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
112 for chr_strand in gtf_gene.keys():
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
113 chr = chr_strand.split("#")[0]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
114 strand = chr_strand.split("#")[1]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
115 start_list = gtf_gene[chr_strand][0]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
116 end_list = gtf_gene[chr_strand][1]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
117 gene_list = gtf_gene[chr_strand][2]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
118 sorted_start_index = [i[0] for i in sorted(enumerate(start_list), key=lambda x:x[1])]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
119 sorted_end_index = [i[0] for i in sorted(enumerate(end_list), key=lambda x:x[1])]
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
120
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
121 sorted_start = sorted(start_list)
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
122 sorted_end = [end_list[i] for i in sorted_start_index]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
123 sorted_gene = [gene_list[i] for i in sorted_start_index]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
124 for x in range(len(sorted_start))[1:]:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
125 intergene_start = sorted_end[x - 1] + 1
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
126 intergene_end = sorted_start[x] - 1
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
127 if intergene_start < intergene_end:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
128 intergene_1 = sorted_gene[x - 1]
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
129 intergene_2 = sorted_gene[x]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
130 gene = intergene_1 + "-#-" + intergene_2
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
131 data = (chr, str(intergene_start), str(intergene_end), gene, "", strand, "intergenic")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
132 gtf[strand]["intergenic"].append(data)
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
133
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
134 import sqlite3
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
135 # conn = sqlite3.connect('gtf_database.db')
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
136 conn = sqlite3.connect(":memory:")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
137 c = conn.cursor()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
138 # c.execute("DROP TABLE IF EXISTS gtf_data;")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
139 # c.execute("CREATE TABLE IF NOT EXISTS gtf_data(chr text, start int, end int, gene text, transcript text, strand text, type text)")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
140 c.execute("CREATE TABLE gtf_data(chr text, start int, end int, gene text, transcript text, strand text, type text)")
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
141
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
142 for strand in gtf.keys():
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
143 if strand not in ["+", "-"]:
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
144 print("Please check the strand information in the GTF file. It should be '+' or '-'.")
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
145
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
146 for type in gtf[strand].keys():
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
147 data = gtf[strand][type]
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
148 c.executemany('INSERT INTO gtf_data VALUES (?,?,?,?,?,?,?)', data)
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
149
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
150 conn.commit()
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
151
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
152 infh = open(inputFile[2], "r")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
153 # infh = open("Mouse_Data_All_peptides_withNewDBs.txt", "r")
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
154 data = infh.readlines()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
155 # output file
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
156 outfh = open(inputFile[3], 'w')
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
157 # outfh = open("classified_1_Mouse_Data_All_peptides_withNewDBs.txt", "w")
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
158
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
159 for each in data:
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
160 a = each.strip().split("\t")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
161 chr = a[0].strip()
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
162 pep_start = str(int(a[1].strip()) + 1)
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
163 pep_end = a[2].strip()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
164 strand = a[5].strip()
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
165 each = "\t".join(a[:6])
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
166 if (len(a) == 12 and int(a[9]) == 1) or (len(a) == 6):
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
167 c.execute("select * from gtf_data where type = 'CDS' and chr = '" + chr + "' and start <= " + pep_start + " and end >= " + pep_end + " and strand = '" + strand + "' ")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
168 rows = c.fetchall()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
169 if len(rows) > 0:
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
170 outfh.write(each.strip() + "\tCDS\n")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
171 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
172 c.execute("select * from gtf_data where type = 'five_prime_utr' and chr = '" + chr + "' and start <= " + pep_start + " and end >= " + pep_end + " and strand = '" + strand + "' ")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
173 rows = c.fetchall()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
174 if len(rows) > 0:
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
175 outfh.write(each.strip() + "\tfive_prime_utr\n")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
176 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
177 c.execute("select * from gtf_data where type = 'three_prime_utr' and chr = '" + chr + "' and start <= " + pep_start + " and end >= " + pep_end + " and strand = '" + strand + "' ")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
178 rows = c.fetchall()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
179 if len(rows) > 0:
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
180 outfh.write(each.strip() + "\tthree_prime_utr\n")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
181 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
182 c.execute("select * from gtf_data where type = 'exon' and chr = '" + chr + "' and start <= " + pep_start + " and end >= " + pep_end + " and strand = '" + strand + "' ")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
183 rows = c.fetchall()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
184 if len(rows) > 0:
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
185 outfh.write(each.strip() + "\texon\n")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
186 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
187 c.execute("select * from gtf_data where type = 'intron' and chr = '" + chr + "' and start <= " + pep_start + " and end >= " + pep_end + " and strand = '" + strand + "' ")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
188 rows = c.fetchall()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
189 if len(rows) > 0:
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
190 outfh.write(each.strip() + "\tintron\n")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
191 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
192 c.execute("select * from gtf_data where type = 'gene' and chr = '" + chr + "' and start <= " + pep_start + " and end >= " + pep_end + " and strand = '" + strand + "' ")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
193 rows = c.fetchall()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
194 if len(rows) > 0:
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
195 outfh.write(each.strip() + "\tgene\n")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
196 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
197 c.execute("select * from gtf_data where type = 'intergenic' and chr = '" + chr + "' and start <= " + pep_start + " and end >= " + pep_end + " and strand = '" + strand + "' ")
2
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
198 rows = c.fetchall()
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
199 if len(rows) > 0:
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
200 outfh.write(each.strip() + "\tintergene\n")
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
201 else:
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
202 outfh.write(each.strip() + "\tOVERLAPPING_ON_TWO_REGIONS: PLEASE_LOOK_MANUALLY (Will be updated in next version)\n")
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
203 elif (len(a) == 12 and int(a[9]) == 2):
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
204 outfh.write(each.strip() + "\tSpliceJunction\n")
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
205 else:
073a2965e3b2 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit e3996c3bda75b16d19997d1e2f67267dd0ea2dff
galaxyp
parents: 0
diff changeset
206 outfh.write(each.strip() + "\tPlease check\n")
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
207
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
208 conn.close()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
209 outfh.close()
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
210 else:
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
211 print("USAGE: python pep_pointer.py <input GTF file> <input tblastn file> <name of output file>")
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
212 return None
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
213
3
a6282baa8c6f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit 494bc6dd87b9a6e2af40cb32aa5d2ee6e9bfebfc
galaxyp
parents: 2
diff changeset
214
0
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
215 if __name__ == "__main__":
149ed6a9680f planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/pep_pointer commit ac27a958fcb897c3cb56db313ebd282805b01103
galaxyp
parents:
diff changeset
216 main()