Mercurial > repos > mbernt > fasta_regex_finder
annotate fastaregexfinder.py @ 1:9a811adb714f draft default tip
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
author | iuc |
---|---|
date | Wed, 25 Jan 2023 14:03:52 +0000 |
parents | 269c627ae9f4 |
children |
rev | line source |
---|---|
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
1 #!/usr/bin/env python |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
2 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
3 import argparse |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
4 import operator |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
5 import re |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
6 import sys |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
7 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
8 VERSION = "0.1.1" |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
9 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
10 parser = argparse.ArgumentParser( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
11 description=""" |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
12 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
13 DESCRIPTION |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
14 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
15 Search a fasta file for matches to a regex and return a bed file with the |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
16 coordinates of the match and the matched sequence itself. |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
17 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
18 Output bed file has columns: |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
19 1. Name of fasta sequence (e.g. chromosome) |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
20 2. Start of the match |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
21 3. End of the match |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
22 4. ID of the match |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
23 5. Length of the match |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
24 6. Strand |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
25 7. Matched sequence as it appears on the forward strand |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
26 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
27 For matches on the reverse strand it is reported the start and end position on the |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
28 forward strand and the matched string on the forward strand (so the G4 'GGGAGGGT' |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
29 present on the reverse strand is reported as ACCCTCCC). |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
30 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
31 Note: Fasta sequences (chroms) are read in memory one at a time along with the |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
32 matches for that chromosome. |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
33 The order of the output is: chroms as they are found in the inut fasta, matches |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
34 sorted within chroms by positions. |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
35 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
36 EXAMPLE: |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
37 ## Test data: |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
38 echo '>mychr' > /tmp/mychr.fa |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
39 echo 'ACTGnACTGnACTGnTGAC' >> /tmp/mychr.fa |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
40 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
41 fastaRegexFinder.py -f /tmp/mychr.fa -r 'ACTG' |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
42 mychr 0 4 mychr_0_4_for 4 + ACTG |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
43 mychr 5 9 mychr_5_9_for 4 + ACTG |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
44 mychr 10 14 mychr_10_14_for 4 + ACTG |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
45 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
46 fastaRegexFinder.py -f /tmp/mychr.fa -r 'ACTG' --maxstr 3 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
47 mychr 0 4 mychr_0_4_for 4 + ACT[3,4] |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
48 mychr 5 9 mychr_5_9_for 4 + ACT[3,4] |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
49 mychr 10 14 mychr_10_14_for 4 + ACT[3,4] |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
50 |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
51 less /tmp/mychr.fa | fastaRegexFinder.py -f - -r 'A\\w\\wGn' |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
52 mychr 0 5 mychr_0_5_for 5 + ACTGn |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
53 mychr 5 10 mychr_5_10_for 5 + ACTGn |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
54 mychr 10 15 mychr_10_15_for 5 + ACTGn |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
55 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
56 DOWNLOAD |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
57 fastaRegexFinder.py is hosted at https://github.com/dariober/bioinformatics-cafe/tree/master/fastaRegexFinder |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
58 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
59 """, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
60 formatter_class=argparse.RawTextHelpFormatter, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
61 ) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
62 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
63 parser.add_argument( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
64 "--fasta", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
65 "-f", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
66 type=str, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
67 help="Input fasta file to search. Use '-' to read the file from stdin.", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
68 required=True, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
69 ) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
70 |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
71 parser.add_argument( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
72 "--regex", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
73 "-r", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
74 type=str, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
75 help="""Regex to be searched in the fasta input. |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
76 Matches to the reverse complement will have - strand. |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
77 The default regex is '([gG]{3,}\\w{1,7}){3,}[gG]{3,}' which searches |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
78 for G-quadruplexes.""", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
79 default="([gG]{3,}\\w{1,7}){3,}[gG]{3,}", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
80 ) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
81 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
82 parser.add_argument( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
83 "--matchcase", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
84 "-m", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
85 action="store_true", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
86 help="""Match case while searching for matches. Default is |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
87 to ignore case (I.e. 'ACTG' will match 'actg'). |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
88 """, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
89 ) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
90 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
91 parser.add_argument( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
92 "--noreverse", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
93 action="store_true", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
94 help="""Do not search the reverse complement of the input fasta. |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
95 Use this flag to search protein sequences.""", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
96 ) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
97 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
98 parser.add_argument( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
99 "--maxstr", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
100 type=int, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
101 required=False, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
102 default=10000, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
103 help="""Maximum length of the match to report in the 7th column of the output. |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
104 Default is to report up to 10000nt. |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
105 Truncated matches are reported as <ACTG...ACTG>[<maxstr>,<tot length>] |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
106 """, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
107 ) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
108 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
109 parser.add_argument( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
110 "--seqnames", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
111 "-s", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
112 type=str, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
113 nargs="+", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
114 default=[None], |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
115 required=False, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
116 help="""List of fasta sequences in --fasta to |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
117 search. E.g. use --seqnames chr1 chr2 chrM to search only these crhomosomes. |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
118 Default is to search all the sequences in input. |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
119 """, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
120 ) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
121 parser.add_argument( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
122 "--quiet", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
123 "-q", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
124 action="store_true", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
125 help="""Do not print progress report (i.e. sequence names as they are scanned). |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
126 """, |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
127 ) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
128 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
129 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
130 parser.add_argument("--version", "-v", action="version", version="%(prog)s " + VERSION) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
131 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
132 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
133 args = parser.parse_args() |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
134 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
135 " --------------------------[ Check and parse arguments ]---------------------- " |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
136 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
137 if args.matchcase: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
138 flag = 0 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
139 else: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
140 flag = re.IGNORECASE |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
141 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
142 " ------------------------------[ Functions ]--------------------------------- " |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
143 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
144 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
145 def sort_table(table, cols): |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
146 """Code to sort list of lists |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
147 see http://www.saltycrane.com/blog/2007/12/how-to-sort-table-by-columns-in-python/ |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
148 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
149 sort a table by multiple columns |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
150 table: a list of lists (or tuple of tuples) where each inner list |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
151 represents a row |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
152 cols: a list (or tuple) specifying the column numbers to sort by |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
153 e.g. (1,0) would sort by column 1, then by column 0 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
154 """ |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
155 for col in reversed(cols): |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
156 table = sorted(table, key=operator.itemgetter(col)) |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
157 return table |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
158 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
159 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
160 def trimMatch(x, n): |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
161 """Trim the string x to be at most length n. Trimmed matches will be reported |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
162 with the syntax ACTG[a,b] where Ns are the beginning of x, a is the length of |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
163 the trimmed strng (e.g 4 here) and b is the full length of the match |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
164 EXAMPLE: |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
165 trimMatch('ACTGNNNN', 4) |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
166 >>>'ACTG[4,8]' |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
167 trimMatch('ACTGNNNN', 8) |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
168 >>>'ACTGNNNN' |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
169 """ |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
170 if len(x) > n and n is not None: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
171 m = x[0:n] + "[" + str(n) + "," + str(len(x)) + "]" |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
172 else: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
173 m = x |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
174 return m |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
175 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
176 |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
177 def revcomp(x): |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
178 """Reverse complement string x. Ambiguity codes are handled and case conserved. |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
179 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
180 Test |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
181 x= 'ACGTRYSWKMBDHVNacgtryswkmbdhvn' |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
182 revcomp(x) |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
183 """ |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
184 compdict = { |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
185 "A": "T", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
186 "C": "G", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
187 "G": "C", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
188 "T": "A", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
189 "R": "Y", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
190 "Y": "R", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
191 "S": "W", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
192 "W": "S", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
193 "K": "M", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
194 "M": "K", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
195 "B": "V", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
196 "D": "H", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
197 "H": "D", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
198 "V": "B", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
199 "N": "N", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
200 "a": "t", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
201 "c": "g", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
202 "g": "c", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
203 "t": "a", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
204 "r": "y", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
205 "y": "r", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
206 "s": "w", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
207 "w": "s", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
208 "k": "m", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
209 "m": "k", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
210 "b": "v", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
211 "d": "h", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
212 "h": "d", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
213 "v": "b", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
214 "n": "n", |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
215 } |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
216 xrc = [] |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
217 for n in x: |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
218 xrc.append(compdict[n]) |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
219 xrc = "".join(xrc)[::-1] |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
220 return xrc |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
221 |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
222 |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
223 # ----------------------------------------------------------------------------- |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
224 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
225 psq_re_f = re.compile(args.regex, flags=flag) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
226 # psq_re_r= re.compile(regexrev) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
227 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
228 if args.fasta != "-": |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
229 ref_seq_fh = open(args.fasta) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
230 else: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
231 ref_seq_fh = sys.stdin |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
232 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
233 ref_seq = [] |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
234 line = (ref_seq_fh.readline()).strip() |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
235 chr = re.sub("^>", "", line) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
236 line = (ref_seq_fh.readline()).strip() |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
237 gquad_list = [] |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
238 while True: |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
239 if not args.quiet: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
240 sys.stderr.write("Processing %s\n" % (chr)) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
241 while line.startswith(">") is False: |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
242 ref_seq.append(line) |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
243 line = (ref_seq_fh.readline()).strip() |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
244 if line == "": |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
245 break |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
246 ref_seq = "".join(ref_seq) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
247 if args.seqnames == [None] or chr in args.seqnames: |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
248 for m in re.finditer(psq_re_f, ref_seq): |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
249 matchstr = trimMatch(m.group(0), args.maxstr) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
250 quad_id = str(chr) + "_" + str(m.start()) + "_" + str(m.end()) + "_for" |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
251 gquad_list.append( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
252 [chr, m.start(), m.end(), quad_id, len(m.group(0)), "+", matchstr] |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
253 ) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
254 if args.noreverse is False: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
255 ref_seq = revcomp(ref_seq) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
256 seqlen = len(ref_seq) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
257 for m in re.finditer(psq_re_f, ref_seq): |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
258 matchstr = trimMatch(revcomp(m.group(0)), args.maxstr) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
259 mstart = seqlen - m.end() |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
260 mend = seqlen - m.start() |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
261 quad_id = str(chr) + "_" + str(mstart) + "_" + str(mend) + "_rev" |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
262 gquad_list.append( |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
263 [chr, mstart, mend, quad_id, len(m.group(0)), "-", matchstr] |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
264 ) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
265 gquad_sorted = sort_table(gquad_list, (1, 2, 3)) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
266 gquad_list = [] |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
267 for xline in gquad_sorted: |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
268 xline = "\t".join([str(x) for x in xline]) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
269 print(xline) |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
270 chr = re.sub("^>", "", line) |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
271 ref_seq = [] |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
272 line = (ref_seq_fh.readline()).strip() |
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
273 if line == "": |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
274 break |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
275 |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
276 # gquad_sorted= sort_table(gquad_list, (0,1,2,3)) |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
277 # |
1
9a811adb714f
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/main/tools/bioinformatics-cafe commit c318cd3f894f89c13d7eb257678673da70f72212
iuc
parents:
0
diff
changeset
|
278 # for line in gquad_sorted: |
0
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
279 # line= '\t'.join([str(x) for x in line]) |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
280 # print(line) |
269c627ae9f4
planemo upload for repository https://github.com/bernt-matthias/mb-galaxy-tools/tree/master/tools/fasta_regex_finder commit 8e118a4d24047e2c62912b962e854f789d6ff559
mbernt
parents:
diff
changeset
|
281 sys.exit() |