Mercurial > repos > jankanis > blast2html
comparison blast2html.py @ 73:67b1a319c6dc py2.6
First go at 2.6 compatibility
The tests are running without errors, but they don't produce the same output as python3
author | Jan Kanis <jan.code@jankanis.nl> |
---|---|
date | Wed, 18 Jun 2014 16:21:45 +0200 |
parents | 6ecbfebb9dd9 |
children | 03e044b5bcc2 |
comparison
equal
deleted
inserted
replaced
72:6ecbfebb9dd9 | 73:67b1a319c6dc |
---|---|
1 #!/usr/bin/env python3 | 1 #!/usr/bin/env python3 |
2 # -*- coding: utf-8 -*- | 2 # -*- coding: utf-8 -*- |
3 | 3 |
4 # Actually this program works with both python 2 and 3 | |
5 | |
4 # Copyright The Hyve B.V. 2014 | 6 # Copyright The Hyve B.V. 2014 |
5 # License: GPL version 3 or (at your option) any higher version | 7 # License: GPL version 3 or (at your option) any higher version |
6 | 8 |
7 from __future__ import unicode_literals | 9 from __future__ import unicode_literals, division |
8 | 10 |
9 import sys | 11 import sys |
10 import math | 12 import math |
11 import warnings | 13 import warnings |
14 import six, codecs | |
12 from os import path | 15 from os import path |
13 from itertools import repeat | 16 from itertools import repeat |
14 import argparse | 17 import argparse |
15 from lxml import objectify | 18 from lxml import objectify |
16 import jinja2 | 19 import jinja2 |
17 | 20 |
18 | 21 |
19 | 22 |
20 _filters = {} | 23 _filters = dict() |
21 def filter(func_or_name): | 24 def filter(func_or_name): |
22 "Decorator to register a function as filter in the current jinja environment" | 25 "Decorator to register a function as filter in the current jinja environment" |
23 if isinstance(func_or_name, str): | 26 if isinstance(func_or_name, six.string_types): |
24 def inner(func): | 27 def inner(func): |
25 _filters[func_or_name] = func.__name__ | 28 _filters[func_or_name] = func.__name__ |
26 return func | 29 return func |
27 return inner | 30 return inner |
28 else: | 31 else: |
98 qseq = hsp.Hsp_qseq.text | 101 qseq = hsp.Hsp_qseq.text |
99 midline = hsp.Hsp_midline.text | 102 midline = hsp.Hsp_midline.text |
100 hseq = hsp.Hsp_hseq.text | 103 hseq = hsp.Hsp_hseq.text |
101 | 104 |
102 if not qframe in (1, -1): | 105 if not qframe in (1, -1): |
103 warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_query-frame of {}. (should be 1 or -1)".format(nodeid(hsp), qframe)) | 106 warnings.warn("Error in BlastXML input: Hsp node {0} has a Hsp_query-frame of {1}. (should be 1 or -1)".format(nodeid(hsp), qframe)) |
104 qframe = -1 if qframe < 0 else 1 | 107 qframe = -1 if qframe < 0 else 1 |
105 if not hframe in (1, -1): | 108 if not hframe in (1, -1): |
106 warnings.warn("Error in BlastXML input: Hsp node {} has a Hsp_hit-frame of {}. (should be 1 or -1)".format(nodeid(hsp), hframe)) | 109 warnings.warn("Error in BlastXML input: Hsp node {0} has a Hsp_hit-frame of {1}. (should be 1 or -1)".format(nodeid(hsp), hframe)) |
107 hframe = -1 if hframe < 0 else 1 | 110 hframe = -1 if hframe < 0 else 1 |
108 | 111 |
109 def split(txt): | 112 def split(txt): |
110 return [txt[i:i+linewidth] for i in range(0, len(txt), linewidth)] | 113 return [txt[i:i+linewidth] for i in range(0, len(txt), linewidth)] |
111 | 114 |
112 for qs, mid, hs, offset in zip(split(qseq), split(midline), split(hseq), range(0, len(qseq), linewidth)): | 115 for qs, mid, hs, offset in zip(split(qseq), split(midline), split(hseq), range(0, len(qseq), linewidth)): |
113 yield ( | 116 yield ( |
114 "Query {:>7} {} {}\n".format(qfrom+offset*qframe, qs, qfrom+(offset+len(qs)-1)*qframe) + | 117 "Query {0:>7} {1} {2}\n".format(qfrom+offset*qframe, qs, qfrom+(offset+len(qs)-1)*qframe) + |
115 " {:7} {}\n".format('', mid) + | 118 " {0:7} {1}\n".format('', mid) + |
116 "Subject{:>7} {} {}".format(hfrom+offset*hframe, hs, hfrom+(offset+len(hs)-1)*hframe) | 119 "Subject{0:>7} {1} {2}".format(hfrom+offset*hframe, hs, hfrom+(offset+len(hs)-1)*hframe) |
117 ) | 120 ) |
118 | 121 |
119 if qfrom+(len(qseq)-1)*qframe != qto: | 122 if qfrom+(len(qseq)-1)*qframe != qto: |
120 warnings.warn("Error in BlastXML input: Hsp node {} qseq length mismatch: from {} to {} length {}".format( | 123 warnings.warn("Error in BlastXML input: Hsp node {0} qseq length mismatch: from {1} to {2} length {3}".format( |
121 nodeid(hsp), qfrom, qto, len(qseq))) | 124 nodeid(hsp), qfrom, qto, len(qseq))) |
122 if hfrom+(len(hseq)-1)*hframe != hto: | 125 if hfrom+(len(hseq)-1)*hframe != hto: |
123 warnings.warn("Error in BlastXML input: Hsp node {} hseq length mismatch: from {} to {} length {}".format( | 126 warnings.warn("Error in BlastXML input: Hsp node {0} hseq length mismatch: from {1} to {2} length {3}".format( |
124 nodeid(hsp), hfrom, hto, len(hseq))) | 127 nodeid(hsp), hfrom, hto, len(hseq))) |
125 | 128 |
126 | 129 |
127 | 130 |
128 @filter('len') | 131 @filter('len') |
157 elif frame == -1: | 160 elif frame == -1: |
158 return 'Minus' | 161 return 'Minus' |
159 raise Exception("frame should be either +1 or -1") | 162 raise Exception("frame should be either +1 or -1") |
160 | 163 |
161 def genelink(hit, type='genbank', hsp=None): | 164 def genelink(hit, type='genbank', hsp=None): |
162 if not isinstance(hit, str): | 165 if not isinstance(hit, six.string_types): |
163 hit = hitid(hit) | 166 hit = hitid(hit) |
164 link = "http://www.ncbi.nlm.nih.gov/nucleotide/{}?report={}&log$=nuclalign".format(hit, type) | 167 link = "http://www.ncbi.nlm.nih.gov/nucleotide/{0}?report={1}&log$=nuclalign".format(hit, type) |
165 if hsp != None: | 168 if hsp != None: |
166 link += "&from={}&to={}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to']) | 169 link += "&from={0}&to={1}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to']) |
167 return link | 170 return link |
168 | 171 |
169 | 172 |
170 # javascript escape filter based on Django's, from https://github.com/dsissitka/khan-website/blob/master/templatefilters.py#L112-139 | 173 # javascript escape filter based on Django's, from https://github.com/dsissitka/khan-website/blob/master/templatefilters.py#L112-139 |
171 # I've removed the html escapes, since html escaping is already being performed by the template engine. | 174 # I've removed the html escapes, since html escaping is already being performed by the template engine. |
196 Javascript string literal escape. Note that this only escapes data | 199 Javascript string literal escape. Note that this only escapes data |
197 for embedding within javascript string literals, not in general | 200 for embedding within javascript string literals, not in general |
198 javascript snippets. | 201 javascript snippets. |
199 """ | 202 """ |
200 | 203 |
201 value = str(value) | 204 value = six.text_type(value) |
202 | 205 |
203 for bad, good in _js_escapes: | 206 for bad, good in _js_escapes: |
204 value = value.replace(bad, good) | 207 value = value.replace(bad, good) |
205 | 208 |
206 return value | 209 return value |
223 def __init__(self, input, templatedir, templatename): | 226 def __init__(self, input, templatedir, templatename): |
224 self.input = input | 227 self.input = input |
225 self.templatename = templatename | 228 self.templatename = templatename |
226 | 229 |
227 self.blast = objectify.parse(self.input).getroot() | 230 self.blast = objectify.parse(self.input).getroot() |
228 self.loader = jinja2.FileSystemLoader(searchpath=templatedir) | 231 self.loader = jinja2.FileSystemLoader(searchpath=templatedir, encoding='utf-8') |
229 self.environment = jinja2.Environment(loader=self.loader, | 232 self.environment = jinja2.Environment(loader=self.loader, |
230 lstrip_blocks=True, trim_blocks=True, autoescape=True) | 233 lstrip_blocks=True, trim_blocks=True, autoescape=True) |
231 | 234 |
232 self._addfilters(self.environment) | 235 self._addfilters(self.environment) |
233 | 236 |
247 ('Query length', self.blast["BlastOutput_query-len"]), | 250 ('Query length', self.blast["BlastOutput_query-len"]), |
248 ('Program', self.blast.BlastOutput_version), | 251 ('Program', self.blast.BlastOutput_version), |
249 ('Database', self.blast.BlastOutput_db), | 252 ('Database', self.blast.BlastOutput_db), |
250 ) | 253 ) |
251 | 254 |
252 output.write(template.render(blast=self.blast, | 255 result = template.render(blast=self.blast, |
253 iterations=self.blast.BlastOutput_iterations.Iteration, | 256 iterations=self.blast.BlastOutput_iterations.Iteration, |
254 colors=self.colors, | 257 colors=self.colors, |
255 # match_colors=self.match_colors(), | 258 # match_colors=self.match_colors(), |
256 # hit_info=self.hit_info(), | 259 # hit_info=self.hit_info(), |
257 genelink=genelink, | 260 genelink=genelink, |
258 params=params)) | 261 params=params) |
262 if six.PY2: | |
263 result = result.encode('utf-8') | |
264 output.write(result) | |
259 | 265 |
260 @filter | 266 @filter |
261 def match_colors(self, result): | 267 def match_colors(self, result): |
262 """ | 268 """ |
263 An iterator that yields lists of length-color pairs. | 269 An iterator that yields lists of length-color pairs. |
319 def hsp_val(path): | 325 def hsp_val(path): |
320 return (float(hsp[path]) for hsp in hsps) | 326 return (float(hsp[path]) for hsp in hsps) |
321 | 327 |
322 yield dict(hit = hit, | 328 yield dict(hit = hit, |
323 title = firsttitle(hit), | 329 title = firsttitle(hit), |
324 maxscore = "{:.1f}".format(max(hsp_val('Hsp_bit-score'))), | 330 maxscore = "{0:.1f}".format(max(hsp_val('Hsp_bit-score'))), |
325 totalscore = "{:.1f}".format(sum(hsp_val('Hsp_bit-score'))), | 331 totalscore = "{0:.1f}".format(sum(hsp_val('Hsp_bit-score'))), |
326 cover = "{:.0%}".format(cover_count / query_length), | 332 cover = "{0:.0%}".format(cover_count / query_length), |
327 e_value = "{:.4g}".format(min(hsp_val('Hsp_evalue'))), | 333 e_value = "{0:.4g}".format(min(hsp_val('Hsp_evalue'))), |
328 # FIXME: is this the correct formula vv? | 334 # FIXME: is this the correct formula vv? |
329 ident = "{:.0%}".format(float(min(hsp.Hsp_identity / blastxml_len(hsp) for hsp in hsps))), | 335 ident = "{0:.0%}".format(float(min(hsp.Hsp_identity / blastxml_len(hsp) for hsp in hsps))), |
330 accession = hit.Hit_accession) | 336 accession = hit.Hit_accession) |
331 | 337 |
332 | 338 |
333 def main(): | 339 def main(): |
340 #import pdb; pdb.set_trace() | |
334 default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja') | 341 default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja') |
335 | 342 |
336 parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page", | 343 parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page", |
337 usage="{} [-i] INPUT [-o OUTPUT]".format(sys.argv[0])) | 344 usage="{0} [-i] INPUT [-o OUTPUT]".format(sys.argv[0])) |
338 input_group = parser.add_mutually_exclusive_group(required=True) | 345 input_group = parser.add_mutually_exclusive_group(required=True) |
339 input_group.add_argument('positional_arg', metavar='INPUT', nargs='?', type=argparse.FileType(mode='r'), | 346 input_group.add_argument('positional_arg', metavar='INPUT', nargs='?', type=argparse.FileType(mode='r'), |
340 help='The input Blast XML file, same as -i/--input') | 347 help='The input Blast XML file, same as -i/--input') |
341 input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'), | 348 input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'), |
342 help='The input Blast XML file') | 349 help='The input Blast XML file') |