Mercurial > repos > jankanis > blast2html2

annotate visualise.py @ 7:9e7927673089

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
intermediate commit before converting some tables to divs
author Jan Kanis <jan.code@jankanis.nl>
date Thu, 08 May 2014 16:51:52 +0200
parents 1df2bfce5c24
children 2fbdf2eb27b4
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
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
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
10 from lxml import objectify
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
11 import jinja2
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
12
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 def color_idx(length):
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
15 if length < 40:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
16 return 0
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
17 elif length < 50:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
18 return 1
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
19 elif length < 80:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
20 return 2
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
21 elif length < 200:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
22 return 3
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
23 return 4
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
24
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
25
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
26 colors = ['black', 'blue', 'green', 'magenta', 'red']
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
27
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
28 blast = objectify.parse('blast xml example1.xml').getroot()
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
29 loader = jinja2.FileSystemLoader(searchpath='.')
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
30 environment = jinja2.Environment(loader=loader, lstrip_blocks=True, trim_blocks=True, autoescape=True)
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
31 environment.filters['color'] = lambda length: match_colors[color_idx(length)]
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
32
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
33 query_length = int(blast["BlastOutput_query-len"])
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
34
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
35 hits = blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
36 # sort hits by longest hotspot first
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
37 ordered_hits = sorted(hits,
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
38 key=lambda h: max(hsp['Hsp_align-len'] for hsp in h.Hit_hsps.Hsp),
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
39 reverse=True)
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
40
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
41 def match_colors():
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
42 """
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
43 An iterator that yields lists of length-color pairs.
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
44 """
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
45
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
46 percent_multiplier = 100 / query_length
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
47
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
48 for hit in hits:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
49 # sort hotspots from short to long, so we can overwrite index colors of
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
50 # short matches with those of long ones.
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
51 hotspots = sorted(hit.Hit_hsps.Hsp, key=lambda hsp: hsp['Hsp_align-len'])
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
52 table = bytearray([255]) * query_length
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
53 for hsp in hotspots:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
54 frm = hsp['Hsp_query-from'] - 1
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
55 to = int(hsp['Hsp_query-to'])
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
56 table[frm:to] = repeat(color_idx(hsp['Hsp_align-len']), to - frm)
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
57
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
58 matches = []
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
59 last = table[0]
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
60 count = 0
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
61 for i in range(query_length):
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
62 if table[i] == last:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
63 count += 1
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
64 continue
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
65 matches.append((count * percent_multiplier, colors[last] if last != 255 else 'none'))
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
66 last = table[i]
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
67 count = 1
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
68 matches.append((count * percent_multiplier, colors[last] if last != 255 else 'none'))
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
69
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
70 yield dict(colors=matches, link="#hit"+hit.Hit_num.text, defline=hit.Hit_def)
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
71
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
72
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
73 def queryscale():
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
74 max_labels = 10
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
75 skip = math.ceil(query_length / max_labels)
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
76 percent_multiplier = 100 / query_length
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
77 for i in range(1, query_length+1):
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
78 if i % skip == 0:
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
79 yield dict(label = i, width = skip * percent_multiplier)
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
80 if query_length % skip != 0:
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
81 yield dict(label = query_length, width = (query_length % skip) * percent_multiplier)
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
82
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
83
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
84 def hit_info():
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
85
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
86 for hit in ordered_hits:
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
87 hsps = hit.Hit_hsps.Hsp
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
88
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
89 cover = [False] * query_length
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
90 for hsp in hsps:
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
91 cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, int(hsp['Hsp_align-len']))
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
92 cover_count = cover.count(True)
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
93
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
94 def hsp_val(path):
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
95 return (hsp[path] for hsp in hsps)
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
96
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
97 yield dict(description = hit.Hit_def,
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
98 maxscore = max(hsp_val('Hsp_bit-score')),
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
99 totalscore = sum(hsp_val('Hsp_bit-score')),
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
100 cover = "{:.0%}".format(cover_count / query_length),
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
101 e_value = min(hsp_val('Hsp_evalue')),
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
102 # FIXME: is this the correct formula vv?
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
103 ident = "{:.0%}".format(min(hsp.Hsp_identity / hsp['Hsp_align-len'] for hsp in hsps)),
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
104 accession = hit.Hit_accession)
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
105
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
106
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
107 def main():
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
108 template = environment.get_template('visualise.html.jinja')
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
109
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
110 params = (('Query ID', blast["BlastOutput_query-ID"]),
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
111 ('Query definition', blast["BlastOutput_query-def"]),
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
112 ('Query length', blast["BlastOutput_query-len"]),
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
113 ('Program', blast.BlastOutput_version),
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
114 ('Database', blast.BlastOutput_db),
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
115 )
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
116
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
117 if len(blast.BlastOutput_iterations.Iteration) > 1:
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
118 warnings.warn("Multiple 'Iteration' elements found, showing only the first")
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
119
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
120 sys.stdout.write(template.render(blast=blast,
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
121 length=query_length,
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
122 #hits=blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit,
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
123 colors=colors,
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
124 match_colors=match_colors(),
7
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
125 queryscale=queryscale(),
9e7927673089 intermediate commit before converting some tables to divs
Jan Kanis <jan.code@jankanis.nl>
parents: 5
diff changeset
126 hit_info=hit_info(),
5
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
127 params=params))
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
128
1df2bfce5c24 first features are working, partial match table
Jan Kanis <jan.code@jankanis.nl>
parents:
diff changeset
129 main()