blast2html: blast2html.py annotate

annotate 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

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
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	4 # Actually this program works with both python 2 and 3
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	5
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	6 # Copyright The Hyve B.V. 2014
72 6ecbfebb9dd9 doc changes Jan Kanis <jan.code@jankanis.nl> parents: 71 diff changeset	7 # License: GPL version 3 or (at your option) any higher version
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	8
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	9 from __future__ import unicode_literals, division
50 bfc82a8aa3c9 unicodify python sources Jan Kanis <jan.code@jankanis.nl> parents: 27 diff changeset	10
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	11 import sys
7 9e7927673089 intermediate commit before converting some tables to divs Jan Kanis <jan.code@jankanis.nl> parents: 5 diff changeset	12 import math
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	13 import warnings
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	14 import six, codecs
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	15 from os import path
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	16 from itertools import repeat
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	17 import argparse
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	18 from lxml import objectify
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	19 import jinja2
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	20
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	21
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	22
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	23 _filters = dict()
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	24 def filter(func_or_name):
11 7660519f2dc9 proper layout for alignments, added some links Jan Kanis <jan.code@jankanis.nl> parents: 10 diff changeset	25 "Decorator to register a function as filter in the current jinja environment"
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	26 if isinstance(func_or_name, six.string_types):
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	27 def inner(func):
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	28 _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	29 return func
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	30 return inner
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	31 else:
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	32 _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	33 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	34
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	35
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	36 def color_idx(length):
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	37 if length < 40:
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	38 return 0
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	39 elif length < 50:
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	40 return 1
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	41 elif length < 80:
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	42 return 2
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	43 elif length < 200:
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	44 return 3
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	45 return 4
1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	46
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	47 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	48 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	49 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	50
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	51 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	52 def firsttitle(hit):
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	53 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	54
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	55 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	56 def othertitles(hit):
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	57 """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	58 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	59
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	60 titles = []
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	61 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	62 fullid, title = t.split(' ', 1)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	63 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	64 titles.append(dict(id = id,
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	65 hitid = hitid,
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	66 fullid = fullid,
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	67 title = title))
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	68 return titles
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	69
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	70 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	71 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	72 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	73 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	74 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	75 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	76 return hitid
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	77
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	78 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	79 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	80 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	81 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	82 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	83 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	84 return hitid
c8347745bbad use Iteration_message tag; also work with unexpected Hit_id values Jan Kanis <jan.code@jankanis.nl> parents: 23 diff changeset	85
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	86
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	87 @filter
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	88 def alignment_pre(hsp):
68 fa8a93bdefd7 fix bug in calculations of alignment end Jan Kanis <jan.code@jankanis.nl> parents: 67 diff changeset	89 """Create the preformatted alignment blocks"""
71 371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	90
371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	91 # line break length
371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	92 linewidth = 60
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	93
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	94 qfrom = int(hsp['Hsp_query-from'])
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	95 qto = int(hsp['Hsp_query-to'])
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	96 qframe = int(hsp['Hsp_query-frame'])
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	97 hfrom = int(hsp['Hsp_hit-from'])
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	98 hto = int(hsp['Hsp_hit-to'])
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	99 hframe = int(hsp['Hsp_hit-frame'])
71 371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	100
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	101 qseq = hsp.Hsp_qseq.text
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	102 midline = hsp.Hsp_midline.text
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	103 hseq = hsp.Hsp_hseq.text
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	104
71 371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	105 if not qframe in (1, -1):
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	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))
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	107 qframe = -1 if qframe < 0 else 1
71 371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	108 if not hframe in (1, -1):
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	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))
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	110 hframe = -1 if hframe < 0 else 1
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	111
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	112 def split(txt):
71 371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	113 return [txt[i:i+linewidth] for i in range(0, len(txt), linewidth)]
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	114
71 371cd585e459 refactor Jan Kanis <jan.code@jankanis.nl> parents: 69 diff changeset	115 for qs, mid, hs, offset in zip(split(qseq), split(midline), split(hseq), range(0, len(qseq), linewidth)):
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	116 yield (
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	117 "Query {0:>7} {1} {2}\n".format(qfrom+offsetqframe, qs, qfrom+(offset+len(qs)-1)qframe) +
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	118 " {0:7} {1}\n".format('', mid) +
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	119 "Subject{0:>7} {1} {2}".format(hfrom+offsethframe, hs, hfrom+(offset+len(hs)-1)hframe)
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	120 )
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	121
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	122 if qfrom+(len(qseq)-1)*qframe != qto:
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	123 warnings.warn("Error in BlastXML input: Hsp node {0} qseq length mismatch: from {1} to {2} length {3}".format(
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	124 nodeid(hsp), qfrom, qto, len(qseq)))
69 0c4ac210068b handle reverse matches Jan Kanis <jan.code@jankanis.nl> parents: 68 diff changeset	125 if hfrom+(len(hseq)-1)*hframe != hto:
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	126 warnings.warn("Error in BlastXML input: Hsp node {0} hseq length mismatch: from {1} to {2} length {3}".format(
67 19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	127 nodeid(hsp), hfrom, hto, len(hseq)))
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	128
19c48f2ec775 wrap alignments if they are too long Jan Kanis <jan.code@jankanis.nl> parents: 53 diff changeset	129
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	130
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	131 @filter('len')
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	132 def blastxml_len(node):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	133 if node.tag == 'Hsp':
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	134 return int(node['Hsp_align-len'])
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	135 elif node.tag == 'Iteration':
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	136 return int(node['Iteration_query-len'])
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	137 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	138
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	139 @filter
53 4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	140 def nodeid(node):
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	141 id = []
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	142 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	143 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	144 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	145 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	146 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	147 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	148 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	149 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	150 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	151 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	152 return '-'.join(id)
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	153 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	154
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	155
4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	156 @filter
10 2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	157 def asframe(frame):
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	158 if frame == 1:
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	159 return 'Plus'
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	160 elif frame == -1:
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	161 return 'Minus'
2fbdf2eb27b4 All data is displayed now, still some formatting to do Jan Kanis <jan.code@jankanis.nl> parents: 7 diff changeset	162 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	163
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	164 def genelink(hit, type='genbank', hsp=None):
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	165 if not isinstance(hit, six.string_types):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	166 hit = hitid(hit)
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	167 link = "http://www.ncbi.nlm.nih.gov/nucleotide/{0}?report={1}&log$=nuclalign".format(hit, type)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	168 if hsp != None:
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	169 link += "&from={0}&to={1}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to'])
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	170 return link
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	171
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	172
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	173 # 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	174 # 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	175
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	176 _base_js_escapes = (
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	177 ('\\', r'\u005C'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	178 ('\'', r'\u0027'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	179 ('"', r'\u0022'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	180 # ('>', r'\u003E'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	181 # ('<', r'\u003C'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	182 # ('&', r'\u0026'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	183 # ('=', r'\u003D'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	184 # ('-', r'\u002D'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	185 # (';', r'\u003B'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	186 # (u'\u2028', r'\u2028'),
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	187 # (u'\u2029', r'\u2029')
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	188 )
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	189
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	190 # 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	191 # 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	192 # parsing mode.
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	193 _js_escapes = (_base_js_escapes +
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	194 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	195
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	196 @filter
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	197 def js_string_escape(value):
72 6ecbfebb9dd9 doc changes Jan Kanis <jan.code@jankanis.nl> parents: 71 diff changeset	198 """
6ecbfebb9dd9 doc changes Jan Kanis <jan.code@jankanis.nl> parents: 71 diff changeset	199 Javascript string literal escape. Note that this only escapes data
6ecbfebb9dd9 doc changes Jan Kanis <jan.code@jankanis.nl> parents: 71 diff changeset	200 for embedding within javascript string literals, not in general
6ecbfebb9dd9 doc changes Jan Kanis <jan.code@jankanis.nl> parents: 71 diff changeset	201 javascript snippets.
6ecbfebb9dd9 doc changes Jan Kanis <jan.code@jankanis.nl> parents: 71 diff changeset	202 """
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	203
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	204 value = six.text_type(value)
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	205
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	206 for bad, good in _js_escapes:
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	207 value = value.replace(bad, good)
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	208
db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	209 return value
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	210
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	211 @filter
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	212 def hits(result):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	213 # sort hits by longest hotspot first
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	214 return sorted(result.Iteration_hits.findall('Hit'),
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	215 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	216 reverse=True)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	217
7 9e7927673089 intermediate commit before converting some tables to divs Jan Kanis <jan.code@jankanis.nl> parents: 5 diff changeset	218
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	219
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	220 class BlastVisualize:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	221
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	222 colors = ('black', 'blue', 'green', 'magenta', 'red')
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	223
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	224 max_scale_labels = 10
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	225
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	226 def __init__(self, input, templatedir, templatename):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	227 self.input = input
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	228 self.templatename = templatename
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	229
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	230 self.blast = objectify.parse(self.input).getroot()
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	231 self.loader = jinja2.FileSystemLoader(searchpath=templatedir, encoding='utf-8')
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	232 self.environment = jinja2.Environment(loader=self.loader,
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	233 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	234
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	235 self._addfilters(self.environment)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	236
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	237
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	238 def _addfilters(self, environment):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	239 for filtername, funcname in _filters.items():
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	240 try:
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	241 environment.filters[filtername] = getattr(self, funcname)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	242 except AttributeError:
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	243 environment.filters[filtername] = globals()[funcname]
12 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 def render(self, output):
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	246 template = self.environment.get_template(self.templatename)
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	247
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	248 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	249 ('Query definition', self.blast["BlastOutput_query-def"]),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	250 ('Query length', self.blast["BlastOutput_query-len"]),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	251 ('Program', self.blast.BlastOutput_version),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	252 ('Database', self.blast.BlastOutput_db),
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	253 )
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	254
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	255 result = template.render(blast=self.blast,
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	256 iterations=self.blast.BlastOutput_iterations.Iteration,
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	257 colors=self.colors,
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	258 # match_colors=self.match_colors(),
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	259 # hit_info=self.hit_info(),
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	260 genelink=genelink,
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	261 params=params)
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	262 if six.PY2:
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	263 result = result.encode('utf-8')
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	264 output.write(result)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	265
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	266 @filter
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	267 def match_colors(self, result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	268 """
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	269 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	270 """
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	271
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	272 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	273
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	274 percent_multiplier = 100 / query_length
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	275
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	276 for hit in hits(result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	277 # 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	278 # 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	279 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	280 table = bytearray([255]) * query_length
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	281 for hsp in hotspots:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	282 frm = hsp['Hsp_query-from'] - 1
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	283 to = int(hsp['Hsp_query-to'])
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	284 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	285
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	286 matches = []
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	287 last = table[0]
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	288 count = 0
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	289 for i in range(query_length):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	290 if table[i] == last:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	291 count += 1
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	292 continue
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	293 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	294 last = table[i]
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	295 count = 1
16 db7e4ee3be03 fix validation, reindent Jan Kanis <jan.code@jankanis.nl> parents: 15 diff changeset	296 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	297
53 4217bb9cf1d3 depend on python 3; fix internal links with multiple iterations Jan Kanis <jan.code@jankanis.nl> parents: 51 diff changeset	298 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	299
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	300 @filter
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	301 def queryscale(self, result):
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	302 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	303 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	304 percent_multiplier = 100 / query_length
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	305 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	306 if i % skip == 0:
23 6995a6f34f3f fix sectioning, add footer Jan Kanis <jan.code@jankanis.nl> parents: 22 diff changeset	307 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	308 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	309 yield dict(label = query_length,
23 6995a6f34f3f fix sectioning, add footer Jan Kanis <jan.code@jankanis.nl> parents: 22 diff changeset	310 width = (query_length % skip) * percent_multiplier)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	311
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	312 @filter
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	313 def hit_info(self, result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	314
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	315 query_length = blastxml_len(result)
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	316
67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	317 for hit in hits(result):
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	318 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	319
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	320 cover = [False] * query_length
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	321 for hsp in hsps:
19 67ddcb807b7d make it work with multiple queries Jan Kanis <jan.code@jankanis.nl> parents: 18 diff changeset	322 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	323 cover_count = cover.count(True)
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	324
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	325 def hsp_val(path):
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	326 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	327
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	328 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	329 title = firsttitle(hit),
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	330 maxscore = "{0:.1f}".format(max(hsp_val('Hsp_bit-score'))),
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	331 totalscore = "{0:.1f}".format(sum(hsp_val('Hsp_bit-score'))),
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	332 cover = "{0:.0%}".format(cover_count / query_length),
67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	333 e_value = "{0:.4g}".format(min(hsp_val('Hsp_evalue'))),
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	334 # FIXME: is this the correct formula vv?
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	335 ident = "{0:.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	336 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	337
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	338
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	339 def main():
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	340 #import pdb; pdb.set_trace()
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	341 default_template = path.join(path.dirname(__file__), 'blast2html.html.jinja')
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	342
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	343 parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page",
73 67b1a319c6dc First go at 2.6 compatibility Jan Kanis <jan.code@jankanis.nl> parents: 72 diff changeset	344 usage="{0} [-i] INPUT [-o OUTPUT]".format(sys.argv[0]))
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	345 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	346 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	347 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	348 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	349 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	350 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	351 help='The output html file')
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	352 # 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	353 # 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	354 # 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	355 # 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	356 # 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	357 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	358 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	359
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	360 args = parser.parse_args()
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	361 if args.input == None:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	362 args.input = args.positional_arg
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	363 if args.input == None:
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	364 parser.error('no input specified')
5 1df2bfce5c24 first features are working, partial match table Jan Kanis <jan.code@jankanis.nl> parents: diff changeset	365
18 4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	366 templatedir, templatename = path.split(args.template.name)
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	367 args.template.close()
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	368 if not templatedir:
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	369 templatedir = '.'
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	370
4434ffab721a add a parameter for the template Jan Kanis <jan.code@jankanis.nl> parents: 16 diff changeset	371 b = BlastVisualize(args.input, templatedir, templatename)
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	372 b.render(args.output)
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	373
11 7660519f2dc9 proper layout for alignments, added some links Jan Kanis <jan.code@jankanis.nl> parents: 10 diff changeset	374
12 a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	375 if __name__ == '__main__':
a459c754cdb5 add links, refactor, proper commandline arguments Jan Kanis <jan.code@jankanis.nl> parents: 11 diff changeset	376 main()
11 7660519f2dc9 proper layout for alignments, added some links Jan Kanis <jan.code@jankanis.nl> parents: 10 diff changeset	377

Mercurial > repos > jankanis > blast2html

annotate blast2html.py @ 73:67b1a319c6dc py2.6