Mercurial > repos > jankanis > blast2html2
annotate blast_html.py @ 16:db7e4ee3be03
fix validation, reindent
author | Jan Kanis <jan.code@jankanis.nl> |
---|---|
date | Tue, 13 May 2014 12:22:35 +0200 |
parents | 648b3b7437da |
children | 4434ffab721a |
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 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
2 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
3 # Copyright The Hyve B.V. 2014 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
4 # License: GPL version 3 or higher |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
5 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
6 import sys |
7
9e7927673089
intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents:
5
diff
changeset
|
7 import math |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
8 import warnings |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
9 from itertools import repeat |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
10 import argparse |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
11 from lxml import objectify |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
12 import jinja2 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
13 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
14 |
10
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
15 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
16 _filters = {} |
10
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
17 def filter(func_or_name): |
11
7660519f2dc9
proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents:
10
diff
changeset
|
18 "Decorator to register a function as filter in the current jinja environment" |
10
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
19 if isinstance(func_or_name, str): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
20 def inner(func): |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
21 _filters[func_or_name] = func |
10
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
22 return func |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
23 return inner |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
24 else: |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
25 _filters[func_or_name.__name__] = func_or_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_or_name |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
27 |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
28 |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
29 def color_idx(length): |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
30 if length < 40: |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
31 return 0 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
32 elif length < 50: |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
33 return 1 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
34 elif length < 80: |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
35 return 2 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
36 elif length < 200: |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
37 return 3 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
38 return 4 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
39 |
10
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
40 @filter |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
41 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
|
42 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
|
43 |
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 firsttitle(hit): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
46 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
|
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 othertitles(hit): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
50 """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
|
51 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
|
52 |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
53 titles = [] |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
54 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
|
55 fullid, title = t.split(' ', 1) |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
56 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
|
57 titles.append(dict(id = id, |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
58 hitid = hitid, |
10
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
59 fullid = fullid, |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
60 title = title)) |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
61 return titles |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
62 |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
63 @filter |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
64 def hitid(hit): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
65 return hit.Hit_id.text.split('|', 2)[1] |
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 seqid(hit): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
69 return hit.Hit_id.text.split('|', 2)[2] |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
70 |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
71 @filter |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
72 def alignment_pre(hsp): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
73 return ( |
15 | 74 "Query {:>7s} {} {}\n".format(hsp['Hsp_query-from'].text, hsp.Hsp_qseq, hsp['Hsp_query-to']) + |
11
7660519f2dc9
proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents:
10
diff
changeset
|
75 " {:7s} {}\n".format('', hsp.Hsp_midline) + |
15 | 76 "Subject{:>7s} {} {}".format(hsp['Hsp_hit-from'].text, hsp.Hsp_hseq, hsp['Hsp_hit-to']) |
11
7660519f2dc9
proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents:
10
diff
changeset
|
77 ) |
10
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
78 |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
79 @filter('len') |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
80 def hsplen(node): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
81 return int(node['Hsp_align-len']) |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
82 |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
83 @filter |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
84 def asframe(frame): |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
85 if frame == 1: |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
86 return 'Plus' |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
87 elif frame == -1: |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
88 return 'Minus' |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
89 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
|
90 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
91 def genelink(hit, type='genbank', hsp=None): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
92 if not isinstance(hit, str): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
93 hit = hitid(hit) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
94 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
|
95 if hsp != None: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
96 link += "&from={}&to={}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to']) |
16 | 97 return link |
98 | |
99 | |
100 # javascript escape filter based on Django's, from https://github.com/dsissitka/khan-website/blob/master/templatefilters.py#L112-139 | |
101 # 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
|
102 |
16 | 103 _base_js_escapes = ( |
104 ('\\', r'\u005C'), | |
105 ('\'', r'\u0027'), | |
106 ('"', r'\u0022'), | |
107 # ('>', r'\u003E'), | |
108 # ('<', r'\u003C'), | |
109 # ('&', r'\u0026'), | |
110 # ('=', r'\u003D'), | |
111 # ('-', r'\u002D'), | |
112 # (';', r'\u003B'), | |
113 # (u'\u2028', r'\u2028'), | |
114 # (u'\u2029', r'\u2029') | |
115 ) | |
116 | |
117 # Escape every ASCII character with a value less than 32. This is | |
118 # needed a.o. to prevent parsers from jumping out of javascript | |
119 # parsing mode. | |
120 _js_escapes = (_base_js_escapes + | |
121 tuple(('%c' % z, '\\u%04X' % z) for z in range(32))) | |
122 | |
123 @filter | |
124 def js_string_escape(value): | |
125 """Escape javascript string literal escapes. Note that this only works | |
126 within javascript string literals, not in general javascript | |
127 snippets.""" | |
128 | |
129 value = str(value) | |
130 | |
131 for bad, good in _js_escapes: | |
132 value = value.replace(bad, good) | |
133 | |
134 return value | |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
135 |
7
9e7927673089
intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents:
5
diff
changeset
|
136 |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
137 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
138 class BlastVisualize: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
139 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
140 colors = ('black', 'blue', 'green', 'magenta', 'red') |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
141 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
142 max_scale_labels = 10 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
143 |
14
0b33898bba45
small bugfixes (one for python 3.4)
Jan Kanis <jan.code@jankanis.nl>
parents:
13
diff
changeset
|
144 templatename = 'blast_html.html.jinja' |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
145 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
146 def __init__(self, input): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
147 self.input = input |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
148 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
149 self.blast = objectify.parse(self.input).getroot() |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
150 self.loader = jinja2.FileSystemLoader(searchpath='.') |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
151 self.environment = jinja2.Environment(loader=self.loader, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
152 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
|
153 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
154 self.environment.filters['color'] = lambda length: match_colors[color_idx(length)] |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
155 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
156 for name, filter in _filters.items(): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
157 self.environment.filters[name] = filter |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
158 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
159 self.query_length = int(self.blast["BlastOutput_query-len"]) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
160 self.hits = self.blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
161 # sort hits by longest hotspot first |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
162 self.ordered_hits = sorted(self.hits, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
163 key=lambda h: max(hsplen(hsp) for hsp in h.Hit_hsps.Hsp), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
164 reverse=True) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
165 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
166 def render(self, output): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
167 template = self.environment.get_template(self.templatename) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
168 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
169 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
|
170 ('Query definition', self.blast["BlastOutput_query-def"]), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
171 ('Query length', self.blast["BlastOutput_query-len"]), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
172 ('Program', self.blast.BlastOutput_version), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
173 ('Database', self.blast.BlastOutput_db), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
174 ) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
175 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
176 if len(self.blast.BlastOutput_iterations.Iteration) > 1: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
177 warnings.warn("Multiple 'Iteration' elements found, showing only the first") |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
178 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
179 output.write(template.render(blast=self.blast, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
180 length=self.query_length, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
181 hits=self.blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
182 colors=self.colors, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
183 match_colors=self.match_colors(), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
184 queryscale=self.queryscale(), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
185 hit_info=self.hit_info(), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
186 genelink=genelink, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
187 params=params)) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
188 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
189 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
190 def match_colors(self): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
191 """ |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
192 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
|
193 """ |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
194 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
195 percent_multiplier = 100 / self.query_length |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
196 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
197 for hit in self.hits: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
198 # 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
|
199 # short matches with those of long ones. |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
200 hotspots = sorted(hit.Hit_hsps.Hsp, key=lambda hsp: hsplen(hsp)) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
201 table = bytearray([255]) * self.query_length |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
202 for hsp in hotspots: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
203 frm = hsp['Hsp_query-from'] - 1 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
204 to = int(hsp['Hsp_query-to']) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
205 table[frm:to] = repeat(color_idx(hsplen(hsp)), to - frm) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
206 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
207 matches = [] |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
208 last = table[0] |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
209 count = 0 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
210 for i in range(self.query_length): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
211 if table[i] == last: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
212 count += 1 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
213 continue |
16 | 214 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
|
215 last = table[i] |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
216 count = 1 |
16 | 217 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
|
218 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
219 yield dict(colors=matches, link="#hit"+hit.Hit_num.text, defline=firsttitle(hit)) |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
220 |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
221 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
222 def queryscale(self): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
223 skip = math.ceil(self.query_length / self.max_scale_labels) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
224 percent_multiplier = 100 / self.query_length |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
225 for i in range(1, self.query_length+1): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
226 if i % skip == 0: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
227 yield dict(label = i, width = skip * percent_multiplier) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
228 if self.query_length % skip != 0: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
229 yield dict(label = self.query_length, width = (self.query_length % skip) * percent_multiplier) |
7
9e7927673089
intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents:
5
diff
changeset
|
230 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
231 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
232 def hit_info(self): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
233 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
234 for hit in self.ordered_hits: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
235 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
|
236 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
237 cover = [False] * self.query_length |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
238 for hsp in hsps: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
239 cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, hsplen(hsp)) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
240 cover_count = cover.count(True) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
241 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
242 def hsp_val(path): |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
243 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
|
244 |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
245 yield dict(hit = hit, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
246 title = firsttitle(hit), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
247 link_id = hit.Hit_num, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
248 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
|
249 totalscore = "{:.1f}".format(sum(hsp_val('Hsp_bit-score'))), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
250 cover = "{:.0%}".format(cover_count / self.query_length), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
251 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
|
252 # FIXME: is this the correct formula vv? |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
253 ident = "{:.0%}".format(float(min(hsp.Hsp_identity / hsplen(hsp) for hsp in hsps))), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
254 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
|
255 |
2fbdf2eb27b4
All data is displayed now, still some formatting to do
Jan Kanis <jan.code@jankanis.nl>
parents:
7
diff
changeset
|
256 |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
257 def main(): |
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
258 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
259 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
|
260 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
|
261 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
|
262 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
|
263 help='The input Blast XML file, same as -i/--input') |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
264 input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'), |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
265 help='The input Blast XML file') |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
266 parser.add_argument('-o', '--output', type=argparse.FileType(mode='w'), default=sys.stdout, |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
267 help='The output html file') |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
268 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
269 args = parser.parse_args() |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
270 if args.input == None: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
271 args.input = args.positional_arg |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
272 if args.input == None: |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
273 parser.error('no input specified') |
5
1df2bfce5c24
first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff
changeset
|
274 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
275 b = BlastVisualize(args.input) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
276 b.render(args.output) |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
277 |
11
7660519f2dc9
proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents:
10
diff
changeset
|
278 |
12
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
279 if __name__ == '__main__': |
a459c754cdb5
add links, refactor, proper commandline arguments
Jan Kanis <jan.code@jankanis.nl>
parents:
11
diff
changeset
|
280 main() |
11
7660519f2dc9
proper layout for alignments, added some links
Jan Kanis <jan.code@jankanis.nl>
parents:
10
diff
changeset
|
281 |