blast2html: blast2html.py annotate

annotate blast2html.py @ 53:4217bb9cf1d3

depend on python 3; fix internal links with multiple iterations

author	Jan Kanis <jan.code@jankanis.nl>
date	Mon, 26 May 2014 13:07:13 +0200
parents	b15a20c2372a
children	19c48f2ec775

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):
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	86 return (
15 648b3b7437da minor update Jan Kanis <jan.code@jankanis.nl> parents: 14 diff changeset	87 "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	88 " {:7s} {}\n".format('', hsp.Hsp_midline) +
15 648b3b7437da minor update Jan Kanis <jan.code@jankanis.nl> parents: 14 diff changeset	89 "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	90 )
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	91
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	92 @filter('len')
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	93 def blastxml_len(node):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	94 if node.tag == 'Hsp':
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	95 return int(node['Hsp_align-len'])
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	96 elif node.tag == 'Iteration':
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	97 return int(node['Iteration_query-len'])
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	98 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	99
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	100 @filter
53 4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	101 def nodeid(node):
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	102 id = []
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	103 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	104 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	105 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	106 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	107 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	108 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	109 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	110 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	111 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	112 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	113 return '-'.join(id)
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	114 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	115
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	116
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	117 @filter
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	118 def asframe(frame):
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	119 if frame == 1:
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	120 return 'Plus'
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	121 elif frame == -1:
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	122 return 'Minus'
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	123 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	124
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	125 def genelink(hit, type='genbank', hsp=None):
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	126 if not isinstance(hit, str):
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	127 hit = hitid(hit)
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	128 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	129 if hsp != None:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	130 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	131 return link
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	132
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	133
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	134 # 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	135 # 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	136
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	137 _base_js_escapes = (
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	138 ('\\', r'\u005C'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	139 ('\'', r'\u0027'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	140 ('"', r'\u0022'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	141 # ('>', r'\u003E'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	142 # ('<', r'\u003C'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	143 # ('&', r'\u0026'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	144 # ('=', r'\u003D'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	145 # ('-', r'\u002D'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	146 # (';', r'\u003B'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	147 # (u'\u2028', r'\u2028'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	148 # (u'\u2029', r'\u2029')
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	149 )
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	150
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	151 # 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	152 # 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	153 # parsing mode.
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	154 _js_escapes = (_base_js_escapes +
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	155 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	156
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	157 @filter
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	158 def js_string_escape(value):
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	159 """Escape javascript string literal escapes. Note that this only works
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	160 within javascript string literals, not in general javascript
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	161 snippets."""
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	162
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	163 value = str(value)
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	164
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	165 for bad, good in _js_escapes:
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	166 value = value.replace(bad, good)
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 return value
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	169
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	170 @filter
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	171 def hits(result):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	172 # sort hits by longest hotspot first
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	173 return sorted(result.Iteration_hits.findall('Hit'),
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	174 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	175 reverse=True)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	176
7 9e7927673089 intermediate commit before converting some tables to divs Jan Kanis <jan.code@jankanis.nl> parents: 5 diff changeset	177
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	178
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	179 class BlastVisualize:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	180
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	181 colors = ('black', 'blue', 'green', 'magenta', 'red')
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	182
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	183 max_scale_labels = 10
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	184
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	185 def __init__(self, input, templatedir, templatename):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	186 self.input = input
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	187 self.templatename = templatename
12 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 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	190 self.loader = jinja2.FileSystemLoader(searchpath=templatedir)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	191 self.environment = jinja2.Environment(loader=self.loader,
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	192 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	193
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	194 self._addfilters(self.environment)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	195
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	196
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	197 def _addfilters(self, environment):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	198 for filtername, funcname in _filters.items():
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	199 try:
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	200 environment.filters[filtername] = getattr(self, funcname)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	201 except AttributeError:
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	202 environment.filters[filtername] = globals()[funcname]
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	203
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	204 def render(self, output):
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	205 template = self.environment.get_template(self.templatename)
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 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	208 ('Query definition', self.blast["BlastOutput_query-def"]),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	209 ('Query length', self.blast["BlastOutput_query-len"]),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	210 ('Program', self.blast.BlastOutput_version),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	211 ('Database', self.blast.BlastOutput_db),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	212 )
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	213
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	214 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	215 iterations=self.blast.BlastOutput_iterations.Iteration,
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	216 colors=self.colors,
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	217 # match_colors=self.match_colors(),
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	218 # hit_info=self.hit_info(),
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	219 genelink=genelink,
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	220 params=params))
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	221
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	222 @filter
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	223 def match_colors(self, result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	224 """
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	225 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	226 """
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 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	229
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	230 percent_multiplier = 100 / query_length
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	231
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	232 for hit in hits(result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	233 # 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	234 # 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	235 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	236 table = bytearray([255]) * query_length
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	237 for hsp in hotspots:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	238 frm = hsp['Hsp_query-from'] - 1
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	239 to = int(hsp['Hsp_query-to'])
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	240 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	241
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	242 matches = []
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	243 last = table[0]
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	244 count = 0
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	245 for i in range(query_length):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	246 if table[i] == last:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	247 count += 1
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	248 continue
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	249 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	250 last = table[i]
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	251 count = 1
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	252 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	253
53 4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	254 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	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 queryscale(self, result):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	258 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	259 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	260 percent_multiplier = 100 / query_length
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	261 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	262 if i % skip == 0:
23 6995a6f34f3f fix sectioning, add footer Jan Kanis <jan.code@jankanis.nl> parents: 22 diff changeset	263 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	264 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	265 yield dict(label = query_length,
23 6995a6f34f3f fix sectioning, add footer Jan Kanis <jan.code@jankanis.nl> parents: 22 diff changeset	266 width = (query_length % skip) * percent_multiplier)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	267
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	268 @filter
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	269 def hit_info(self, result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	270
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	271 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	272
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	273 for hit in hits(result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	274 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	275
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	276 cover = [False] * query_length
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	277 for hsp in hsps:
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	278 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	279 cover_count = cover.count(True)
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	280
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	281 def hsp_val(path):
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	282 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	283
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	284 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	285 title = firsttitle(hit),
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	286 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	287 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	288 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	289 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	290 # 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	291 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	292 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	293
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	294
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	295 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	296 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	297
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	298 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	299 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	300 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	301 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	302 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	303 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	304 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	305 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	306 help='The output html file')
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	307 # 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	308 # 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	309 # 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	310 # 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	311 # 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	312 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	313 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	314
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	315 args = parser.parse_args()
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	316 if args.input == None:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	317 args.input = args.positional_arg
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	318 if args.input == None:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	319 parser.error('no input specified')
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	320
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	321 templatedir, templatename = path.split(args.template.name)
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	322 args.template.close()
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	323 if not templatedir:
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	324 templatedir = '.'
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	325
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	326 b = BlastVisualize(args.input, templatedir, templatename)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	327 b.render(args.output)
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	328
11 7660519f2dc9 proper layout for alignments, added some links Jan Kanis <jan.code@jankanis.nl> parents: 10 diff changeset	329
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	330 if __name__ == '__main__':
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	331 main()
11 7660519f2dc9 proper layout for alignments, added some links Jan Kanis <jan.code@jankanis.nl> parents: 10 diff changeset	332

Mercurial > repos > jankanis > blast2html

annotate blast2html.py @ 53:4217bb9cf1d3