Mercurial > repos > rnateam > rnacommender

annotate pfam_utils/__init__.py @ 0:d04fa5201f51 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
author rnateam
date Thu, 28 Jul 2016 05:56:54 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
1 """Utils for PFAM."""
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
2
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
3 import xml.etree.ElementTree as ET
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
4 from math import ceil
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
5 from time import sleep
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
6 from xml.etree.ElementTree import ParseError
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
7
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
8 import requests
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
9
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
10 import pandas as pd
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
11
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
12 import fasta_utils
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
13
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
14 __author__ = "Gianluca Corrado"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
15 __copyright__ = "Copyright 2016, Gianluca Corrado"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
16 __license__ = "MIT"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
17 __maintainer__ = "Gianluca Corrado"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
18 __email__ = "gianluca.corrado@unitn.it"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
19 __status__ = "Production"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
20
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
21
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
22 def search_header():
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
23 """Return the header of a Pfam scan search."""
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
24 return "<seq id> <alignment start> <alignment end> \
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
25 <envelope start> <envelope end> <hmm acc> <hmm name>\
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
26 <type> <hmm start> <hmm end> <hmm length> <bit score>\
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
27 <E-value> <significance> <clan>\n"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
28
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
29
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
30 def sequence_search(seq_id, seq):
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
31 """
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
32 Search a sequence against PFAM.
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
33
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
34 Input
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
35 -----
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
36 seq_id : str
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
37 Name of the protein sequence.
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
38 seq : str
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
39 Protein sequence.
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
40
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
41 Output
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
42 ------
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
43 ret : str
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
44 Formatted string containing the results of the Pfam scan for the
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
45 given sequence
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
46 """
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
47 def add_spaces(text, mul=8):
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
48 """Add spaces to a string."""
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
49 l = len(text)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
50 next_mul = int(ceil(l / mul) + 1) * mul
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
51 offset = next_mul - l
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
52 if offset == 0:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
53 offset = 8
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
54 return text + " " * offset
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
55
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
56 url = "http://pfam.xfam.org/search/sequence"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
57 params = {'seq': seq,
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
58 'evalue': '1.0',
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
59 'output': 'xml'}
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
60 req = requests.get(url, params=params)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
61 xml = req.text
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
62 try:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
63 root = ET.fromstring(xml)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
64 # sometimes Pfam returns the HTML code
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
65 except ParseError:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
66 print "resending: %s" % seq_id
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
67 return "%s" % sequence_search(seq_id, seq)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
68
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
69 result_url = root[0][1].text
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
70 # wait for Pfam to compute the results
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
71 sleep(4)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
72 while True:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
73 req2 = requests.get(result_url)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
74 if req2.status_code == 200:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
75 break
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
76 else:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
77 sleep(1)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
78 result_xml = req2.text
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
79 root = ET.fromstring(result_xml)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
80 try:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
81 matches = root[0][0][0][0][:]
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
82 # Sometimes raised when the sequence has no matches
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
83 except IndexError:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
84 return ""
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
85 ret = ""
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
86 for match in matches:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
87 for location in match:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
88 ret += add_spaces(seq_id)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
89 ret += add_spaces(location.attrib['ali_start'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
90 ret += add_spaces(location.attrib['ali_end'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
91 ret += add_spaces(location.attrib['start'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
92 ret += add_spaces(location.attrib['end'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
93 ret += add_spaces(match.attrib['accession'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
94 ret += add_spaces(match.attrib['id'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
95 ret += add_spaces(match.attrib['class'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
96 ret += add_spaces(location.attrib['hmm_start'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
97 ret += add_spaces(location.attrib['hmm_end'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
98 ret += add_spaces("None")
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
99 ret += add_spaces(location.attrib['bitscore'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
100 ret += add_spaces(location.attrib['evalue'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
101 ret += add_spaces(location.attrib['significant'])
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
102 ret += "None\n"
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
103 return ret
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
104
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
105
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
106 def read_pfam_output(pfam_out_file):
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
107 """Read the output of PFAM scan."""
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
108 cols = ["seq_id", "alignment_start", "alignment_end", "envelope_start",
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
109 "envelope_end", "hmm_acc", "hmm_name", "type", "hmm_start",
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
110 "hmm_end", "hmm_length", "bit_score", "E-value", "significance",
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
111 "clan"]
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
112 try:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
113 data = pd.read_table(pfam_out_file,
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
114 sep="\s*", skip_blank_lines=True, skiprows=1,
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
115 names=cols, engine='python')
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
116 except:
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
117 return None
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
118 return data
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
119
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
120
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
121 def download_seed_seqs(acc):
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
122 """
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
123 Download seed sequences from PFAM.
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
124
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
125 Input
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
126 -----
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
127 acc : str
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
128 Accession number of a Pfam domain
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
129
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
130 Output
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
131 ------
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
132 fasta : str
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
133 Seed sequences in fasta format
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
134 """
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
135 url = "http://pfam.xfam.org/family/%s/alignment/seed" % acc
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
136 req = requests.get(url)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
137 stockholm = req.text
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
138 fasta = fasta_utils.stockholm2fasta(stockholm)
d04fa5201f51 planemo upload for repository https://github.com/bgruening/galaxytools/tree/rna_commander/tools/rna_tools/rna_commender commit 7ad344d108076116e702e1c1e91cea73d8fcadc4
rnateam
parents:
diff changeset
139 return fasta