annotate blast2html.py @ 69:0c4ac210068b

handle reverse matches
author Jan Kanis <jan.code@jankanis.nl>
date Wed, 18 Jun 2014 14:12:00 +0200
parents fa8a93bdefd7
children 371cd585e459
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
1 #!/usr/bin/env python3
50
bfc82a8aa3c9 unicodify python sources
Jan Kanis <jan.code@jankanis.nl>
parents: 27
diff changeset
2 # -*- coding: utf-8 -*-
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
3
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
4 # Copyright The Hyve B.V. 2014
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
5 # License: GPL version 3 or higher
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
6
50
bfc82a8aa3c9 unicodify python sources
Jan Kanis <jan.code@jankanis.nl>
parents: 27
diff changeset
7 from __future__ import unicode_literals
bfc82a8aa3c9 unicodify python sources
Jan Kanis <jan.code@jankanis.nl>
parents: 27
diff changeset
8
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
9 import sys
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
10 import math
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
11 import warnings
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
12 from os import path
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
13 from itertools import repeat
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
14 import argparse
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
15 from lxml import objectify
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
16 import jinja2
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
17
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
18
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
19
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
20 _filters = {}
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
21 def filter(func_or_name):
11
7660519f2dc9 proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents: 10
diff changeset
22 "Decorator to register a function as filter in the current jinja environment"
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
23 if isinstance(func_or_name, str):
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
24 def inner(func):
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
25 _filters[func_or_name] = func.__name__
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
26 return func
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
27 return inner
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
28 else:
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
29 _filters[func_or_name.__name__] = func_or_name.__name__
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
30 return func_or_name
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
31
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
32
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
33 def color_idx(length):
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
34 if length < 40:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
35 return 0
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
36 elif length < 50:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
37 return 1
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
38 elif length < 80:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
39 return 2
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
40 elif length < 200:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
41 return 3
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
42 return 4
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
43
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
44 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
45 def fmt(val, fmt):
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
46 return format(float(val), fmt)
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
47
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
48 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
49 def firsttitle(hit):
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
50 return hit.Hit_def.text.split('>')[0]
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
51
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
52 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
53 def othertitles(hit):
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
54 """Split a hit.Hit_def that contains multiple titles up, splitting out the hit ids from the titles."""
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
55 id_titles = hit.Hit_def.text.split('>')
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
56
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
57 titles = []
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
58 for t in id_titles[1:]:
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
59 fullid, title = t.split(' ', 1)
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
60 hitid, id = fullid.split('|', 2)[1:3]
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
61 titles.append(dict(id = id,
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
62 hitid = hitid,
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
63 fullid = fullid,
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
64 title = title))
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
65 return titles
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
66
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
67 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
68 def hitid(hit):
24
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
69 hitid = hit.Hit_id.text
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
70 s = hitid.split('|', 2)
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
71 if len(s) >= 2:
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
72 return s[1]
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
73 return hitid
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
74
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
75 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
76 def seqid(hit):
24
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
77 hitid = hit.Hit_id.text
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
78 s = hitid.split('|', 2)
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
79 if len(s) >= 3:
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
80 return s[2]
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
81 return hitid
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values
Jan Kanis <jan.code@jankanis.nl>
parents: 23
diff changeset
82
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
83
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
84 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
85 def alignment_pre(hsp):
68
fa8a93bdefd7 fix bug in calculations of alignment end
Jan Kanis <jan.code@jankanis.nl>
parents: 67
diff changeset
86 """Create the preformatted alignment blocks"""
fa8a93bdefd7 fix bug in calculations of alignment end
Jan Kanis <jan.code@jankanis.nl>
parents: 67
diff changeset
87
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
88 step = 60
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
89
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
90 qfrom = int(hsp['Hsp_query-from'])
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
91 qto = int(hsp['Hsp_query-to'])
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
92 qframe = int(hsp['Hsp_query-frame'])
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
93 hfrom = int(hsp['Hsp_hit-from'])
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
94 hto = int(hsp['Hsp_hit-to'])
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
95 hframe = int(hsp['Hsp_hit-frame'])
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
96 qseq = hsp.Hsp_qseq.text
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
97 midline = hsp.Hsp_midline.text
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
98 hseq = hsp.Hsp_hseq.text
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
99
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
100 if not qframe in [1, -1]:
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
101 warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_query-frame of {}".format(nodeid(hsp), qframe))
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
102 qframe = -1 if qframe < 0 else 1
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
103 if not hframe in [1, -1]:
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
104 warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_hit-frame of {}".format(nodeid(hsp), hframe))
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
105 hframe = -1 if hframe < 0 else 1
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
106
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
107 def split(txt):
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
108 return [txt[i:i+step] for i in range(0, len(txt), step)]
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
109
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
110 for qs, mid, hs, offset in zip(split(qseq), split(midline), split(hseq), range(0, len(qseq), step)):
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
111 yield (
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
112 "Query {:>7} {} {}\n".format(qfrom+offset*qframe, qs, qfrom+(offset+len(qs)-1)*qframe) +
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
113 " {:7} {}\n".format('', mid) +
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
114 "Subject{:>7} {} {}".format(hfrom+offset*hframe, hs, hfrom+(offset+len(hs)-1)*hframe)
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
115 )
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
116
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
117 if qfrom+(len(qseq)-1)*qframe != qto:
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
118 warnings.warn("Error in BlastXML input: Hsp node {} qseq length mismatch: from {} to {} length {}".format(
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
119 nodeid(hsp), qfrom, qto, len(qseq)))
69
0c4ac210068b handle reverse matches
Jan Kanis <jan.code@jankanis.nl>
parents: 68
diff changeset
120 if hfrom+(len(hseq)-1)*hframe != hto:
67
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
121 warnings.warn("Error in BlastXML input: Hsp node {} hseq length mismatch: from {} to {} length {}".format(
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
122 nodeid(hsp), hfrom, hto, len(hseq)))
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
123
19c48f2ec775 wrap alignments if they are too long
Jan Kanis <jan.code@jankanis.nl>
parents: 53
diff changeset
124
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
125
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
126 @filter('len')
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
127 def blastxml_len(node):
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
128 if node.tag == 'Hsp':
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
129 return int(node['Hsp_align-len'])
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
130 elif node.tag == 'Iteration':
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
131 return int(node['Iteration_query-len'])
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
132 raise Exception("Unknown XML node type: "+node.tag)
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
133
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
134 @filter
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
135 def nodeid(node):
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
136 id = []
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
137 if node.tag == 'Hsp':
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
138 id.insert(0, node.Hsp_num.text)
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
139 node = node.getparent().getparent()
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
140 assert node.tag == 'Hit'
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
141 if node.tag == 'Hit':
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
142 id.insert(0, node.Hit_num.text)
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
143 node = node.getparent().getparent()
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
144 assert node.tag == 'Iteration'
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
145 if node.tag == 'Iteration':
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
146 id.insert(0, node['Iteration_iter-num'].text)
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
147 return '-'.join(id)
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
148 raise ValueError("The nodeid filter can only be applied to Hsp, Hit or Iteration nodes in a BlastXML document")
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
149
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
150
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
151 @filter
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
152 def asframe(frame):
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
153 if frame == 1:
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
154 return 'Plus'
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
155 elif frame == -1:
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
156 return 'Minus'
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
157 raise Exception("frame should be either +1 or -1")
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
158
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
159 def genelink(hit, type='genbank', hsp=None):
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
160 if not isinstance(hit, str):
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
161 hit = hitid(hit)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
162 link = "http://www.ncbi.nlm.nih.gov/nucleotide/{}?report={}&log$=nuclalign".format(hit, type)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
163 if hsp != None:
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
164 link += "&from={}&to={}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to'])
16
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
165 return link
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
166
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
167
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
168 # javascript escape filter based on Django's, from https://github.com/dsissitka/khan-website/blob/master/templatefilters.py#L112-139
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
169 # I've removed the html escapes, since html escaping is already being performed by the template engine.
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
170
16
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
171 _base_js_escapes = (
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
172 ('\\', r'\u005C'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
173 ('\'', r'\u0027'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
174 ('"', r'\u0022'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
175 # ('>', r'\u003E'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
176 # ('<', r'\u003C'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
177 # ('&', r'\u0026'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
178 # ('=', r'\u003D'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
179 # ('-', r'\u002D'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
180 # (';', r'\u003B'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
181 # (u'\u2028', r'\u2028'),
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
182 # (u'\u2029', r'\u2029')
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
183 )
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
184
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
185 # Escape every ASCII character with a value less than 32. This is
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
186 # needed a.o. to prevent html parsers from jumping out of javascript
16
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
187 # parsing mode.
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
188 _js_escapes = (_base_js_escapes +
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
189 tuple(('%c' % z, '\\u%04X' % z) for z in range(32)))
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
190
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
191 @filter
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
192 def js_string_escape(value):
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
193 """Escape javascript string literal escapes. Note that this only works
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
194 within javascript string literals, not in general javascript
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
195 snippets."""
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
196
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
197 value = str(value)
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
198
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
199 for bad, good in _js_escapes:
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
200 value = value.replace(bad, good)
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
201
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
202 return value
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
203
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
204 @filter
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
205 def hits(result):
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
206 # sort hits by longest hotspot first
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
207 return sorted(result.Iteration_hits.findall('Hit'),
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
208 key=lambda h: max(blastxml_len(hsp) for hsp in h.Hit_hsps.Hsp),
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
209 reverse=True)
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
210
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
211
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
212
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
213 class BlastVisualize:
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
214
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
215 colors = ('black', 'blue', 'green', 'magenta', 'red')
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
216
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
217 max_scale_labels = 10
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
218
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
219 def __init__(self, input, templatedir, templatename):
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
220 self.input = input
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
221 self.templatename = templatename
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
222
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
223 self.blast = objectify.parse(self.input).getroot()
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
224 self.loader = jinja2.FileSystemLoader(searchpath=templatedir)
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
225 self.environment = jinja2.Environment(loader=self.loader,
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
226 lstrip_blocks=True, trim_blocks=True, autoescape=True)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
227
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
228 self._addfilters(self.environment)
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
229
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
230
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
231 def _addfilters(self, environment):
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
232 for filtername, funcname in _filters.items():
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
233 try:
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
234 environment.filters[filtername] = getattr(self, funcname)
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
235 except AttributeError:
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
236 environment.filters[filtername] = globals()[funcname]
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
237
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
238 def render(self, output):
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
239 template = self.environment.get_template(self.templatename)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
240
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
241 params = (('Query ID', self.blast["BlastOutput_query-ID"]),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
242 ('Query definition', self.blast["BlastOutput_query-def"]),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
243 ('Query length', self.blast["BlastOutput_query-len"]),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
244 ('Program', self.blast.BlastOutput_version),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
245 ('Database', self.blast.BlastOutput_db),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
246 )
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
247
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
248 output.write(template.render(blast=self.blast,
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
249 iterations=self.blast.BlastOutput_iterations.Iteration,
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
250 colors=self.colors,
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
251 # match_colors=self.match_colors(),
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
252 # hit_info=self.hit_info(),
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
253 genelink=genelink,
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
254 params=params))
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
255
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
256 @filter
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
257 def match_colors(self, result):
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
258 """
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
259 An iterator that yields lists of length-color pairs.
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
260 """
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
261
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
262 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
263
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
264 percent_multiplier = 100 / query_length
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
265
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
266 for hit in hits(result):
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
267 # sort hotspots from short to long, so we can overwrite index colors of
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
268 # short matches with those of long ones.
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
269 hotspots = sorted(hit.Hit_hsps.Hsp, key=lambda hsp: blastxml_len(hsp))
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
270 table = bytearray([255]) * query_length
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
271 for hsp in hotspots:
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
272 frm = hsp['Hsp_query-from'] - 1
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
273 to = int(hsp['Hsp_query-to'])
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
274 table[frm:to] = repeat(color_idx(blastxml_len(hsp)), to - frm)
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
275
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
276 matches = []
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
277 last = table[0]
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
278 count = 0
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
279 for i in range(query_length):
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
280 if table[i] == last:
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
281 count += 1
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
282 continue
16
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
283 matches.append((count * percent_multiplier, self.colors[last] if last != 255 else 'transparent'))
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
284 last = table[i]
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
285 count = 1
16
db7e4ee3be03 fix validation, reindent
Jan Kanis <jan.code@jankanis.nl>
parents: 15
diff changeset
286 matches.append((count * percent_multiplier, self.colors[last] if last != 255 else 'transparent'))
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
287
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
288 yield dict(colors=matches, hit=hit, defline=firsttitle(hit))
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
289
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
290 @filter
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
291 def queryscale(self, result):
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
292 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
293 skip = math.ceil(query_length / self.max_scale_labels)
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
294 percent_multiplier = 100 / query_length
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
295 for i in range(1, query_length+1):
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
296 if i % skip == 0:
23
6995a6f34f3f fix sectioning, add footer
Jan Kanis <jan.code@jankanis.nl>
parents: 22
diff changeset
297 yield dict(label = i, width = skip * percent_multiplier)
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
298 if query_length % skip != 0:
20
53cd304c5f26 Add index for multiple results; fix layout of query ruler for edge case
Jan Kanis <jan.code@jankanis.nl>
parents: 19
diff changeset
299 yield dict(label = query_length,
23
6995a6f34f3f fix sectioning, add footer
Jan Kanis <jan.code@jankanis.nl>
parents: 22
diff changeset
300 width = (query_length % skip) * percent_multiplier)
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
301
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
302 @filter
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
303 def hit_info(self, result):
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
304
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
305 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
306
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
307 for hit in hits(result):
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
308 hsps = hit.Hit_hsps.Hsp
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
309
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
310 cover = [False] * query_length
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
311 for hsp in hsps:
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
312 cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, blastxml_len(hsp))
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
313 cover_count = cover.count(True)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
314
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
315 def hsp_val(path):
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
316 return (float(hsp[path]) for hsp in hsps)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
317
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
318 yield dict(hit = hit,
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
319 title = firsttitle(hit),
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
320 maxscore = "{:.1f}".format(max(hsp_val('Hsp_bit-score'))),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
321 totalscore = "{:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
322 cover = "{:.0%}".format(cover_count / query_length),
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
323 e_value = "{:.4g}".format(min(hsp_val('Hsp_evalue'))),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
324 # FIXME: is this the correct formula vv?
19
67ddcb807b7d make it work with multiple queries
Jan Kanis <jan.code@jankanis.nl>
parents: 18
diff changeset
325 ident = "{:.0%}".format(float(min(hsp.Hsp_identity / blastxml_len(hsp) for hsp in hsps))),
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
326 accession = hit.Hit_accession)
10
2fbdf2eb27b4 All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents: 7
diff changeset
327
27
4e6ac737ba17 improve the galaxy html stripping warning; make sure the tool can find the template from within galaxy
Jan Kanis <jan.code@jankanis.nl>
parents: 24
diff changeset
328
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
329 def main():
27
4e6ac737ba17 improve the galaxy html stripping warning; make sure the tool can find the template from within galaxy
Jan Kanis <jan.code@jankanis.nl>
parents: 24
diff changeset
330 default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja')
4e6ac737ba17 improve the galaxy html stripping warning; make sure the tool can find the template from within galaxy
Jan Kanis <jan.code@jankanis.nl>
parents: 24
diff changeset
331
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
332 parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page",
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
333 usage="{} [-i] INPUT [-o OUTPUT]".format(sys.argv[0]))
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
334 input_group = parser.add_mutually_exclusive_group(required=True)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
335 input_group.add_argument('positional_arg', metavar='INPUT', nargs='?', type=argparse.FileType(mode='r'),
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
336 help='The input Blast XML file, same as -i/--input')
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
337 input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'),
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
338 help='The input Blast XML file')
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
339 parser.add_argument('-o', '--output', type=argparse.FileType(mode='w'), default=sys.stdout,
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
340 help='The output html file')
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
341 # We just want the file name here, so jinja can open the file
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
342 # itself. But it is easier to just use a FileType so argparse can
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
343 # handle the errors. This introduces a small race condition when
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
344 # jinja later tries to re-open the template file, but we don't
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
345 # care too much.
53
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations
Jan Kanis <jan.code@jankanis.nl>
parents: 51
diff changeset
346 parser.add_argument('--template', type=argparse.FileType(mode='r'), default=default_template,
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
347 help='The template file to use. Defaults to blast_html.html.jinja')
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
348
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
349 args = parser.parse_args()
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
350 if args.input == None:
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
351 args.input = args.positional_arg
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
352 if args.input == None:
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
353 parser.error('no input specified')
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
354
18
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
355 templatedir, templatename = path.split(args.template.name)
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
356 args.template.close()
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
357 if not templatedir:
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
358 templatedir = '.'
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
359
4434ffab721a add a parameter for the template
Jan Kanis <jan.code@jankanis.nl>
parents: 16
diff changeset
360 b = BlastVisualize(args.input, templatedir, templatename)
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
361 b.render(args.output)
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
362
11
7660519f2dc9 proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents: 10
diff changeset
363
12
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
364 if __name__ == '__main__':
a459c754cdb5 add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents: 11
diff changeset
365 main()
11
7660519f2dc9 proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents: 10
diff changeset
366