Mercurial > repos > jankanis > blast2html
changeset 12:a459c754cdb5
add links, refactor, proper commandline arguments
| author | Jan Kanis <jan.code@jankanis.nl> |
|---|---|
| date | Mon, 12 May 2014 17:12:24 +0200 |
| parents | 7660519f2dc9 |
| children | c2d63adb83db |
| files | visualise.html.jinja visualise.py |
| diffstat | 2 files changed, 196 insertions(+), 124 deletions(-) [+] |
line wrap: on
line diff
--- a/visualise.html.jinja Mon May 12 13:55:04 2014 +0200 +++ b/visualise.html.jinja Mon May 12 17:12:24 2014 +0200 @@ -11,6 +11,14 @@ font-family: Arial,Sans-Serif; } + :link { + color: #336699; + } + + .right { + float: right; + } + #content { margin: 0 2em; padding: 0.5em; @@ -76,10 +84,12 @@ float: left; } + + .graphic { background-color: white; border: 2px solid black; - padding: .5em; + padding: .5em 1.5em; align: center; margin: auto; } @@ -128,7 +138,7 @@ } /* For small widths we give the graphic 100% */ - @media (max-width: 70em) { + @media (max-width: 72.5em) { div.tablewrapper { width: 100%; min-width: 0px; @@ -174,6 +184,8 @@ height: 4px; } + + div#descriptions { } @@ -212,10 +224,9 @@ white-space: nowrap; text-align: left; } - table.descriptiontable a:link { - color: #336699; - } + + #alignments .white { padding: 1.5em 1em; } @@ -228,12 +239,13 @@ div.linkheader { padding-top: .2em; - text-align: right; font-size: 85%; color: #14376C; } - div.linkheader a { - color: #14376C; + div.linkheader a.linkheader { + margin-right: 1em; + } + div.linkheader .right a { text-decoration: none; } @@ -244,6 +256,7 @@ .title .titleinfo { font-size: 80%; margin-top: 0; + margin-bottom: .3em; } .title .titleinfo .b { color: #606060; @@ -259,13 +272,15 @@ padding: 0; } .moretitles .hittitle { - margin: .2em 0; + margin: .4em 0 .2em 0; padding: 0; } a.showmoretitles { - font-size: 85%; + font-size: 75%; color: #336699; + font-weight: bold; + margin-top: 0; } a.showmoretitles:hover { } @@ -276,13 +291,18 @@ margin-bottom: 2.5em; } - .hotspot .range { + .hotspot p.range { font-size: 70%; margin-top: 0; - font-weight: bold; margin-top: 1em; margin-bottom: .2em; } + .hotspot p.range span.range { + font-weight: bold; + } + .hotspot p.range a.range { + margin-left: .5em; + } table.hotspotstable { border-top: 1px solid; @@ -330,14 +350,14 @@ <table class=headerdata> {% for param, value in params %} - <tr><td class=param>{{param}}</td><td>{{value}}</td></tr> + <tr><td class=param>{{param}}:</td><td>{{value}}</td></tr> {% endfor %} </table> </div> - - {% if not (blast.BlastOutput_iterations.find('Iteration') and - blast.BlastOutput_iterations.Iteration.Iteration_hits.find('Hit')) %} + + {% if not (blast.BlastOutput_iterations.findall('Iteration') and + blast.BlastOutput_iterations.Iteration.Iteration_hits.findall('Hit')) %} <div id=nodata> <h2>No Results</h2> <div class=grey> @@ -360,9 +380,9 @@ <h4 class=darkHeader>Color key for alignment scores</h4> <div class=legend><div class=graphicrow> <div class=graphicitem style="background-color: {{colors[0]}}"><40</div> - <div class=graphicitem style="background-color: {{colors[1]}}">40&endash;50</div> - <div class=graphicitem style="background-color: {{colors[2]}}">50-80</div> - <div class=graphicitem style="background-color: {{colors[3]}}">80-200</div> + <div class=graphicitem style="background-color: {{colors[1]}}">40–50</div> + <div class=graphicitem style="background-color: {{colors[2]}}">50–80</div> + <div class=graphicitem style="background-color: {{colors[3]}}">80–200</div> <div class=graphicitem style="background-color: {{colors[4]}}">200≤</div> </div></div> <div style="clear: left"></div> @@ -433,7 +453,7 @@ <td>{{hit.cover}}</td> <td>{{hit.e_value}}</td> <td>{{hit.ident}}</td> - <td><a href="http://www.ncbi.nlm.nih.gov/nucleotide/{{hit.hit|hitid}}?report=genbank&log$=nucltop&blast_rank=1">{{hit.accession}}</a></td> + <td><a href="{{genelink(hit.hit|hitid)}}">{{hit.accession}}</a></td> </tr> {% endfor %} </table> @@ -451,13 +471,15 @@ <div class=alignment id=hit{{hit.Hit_num}}> <div class=linkheader> - <a href="#description{{hit.Hit_num}}">Descriptions</a> + <div class=right><a href="#description{{hit.Hit_num}}">Descriptions</a></div> + <a class=linkheader href="{{genelink(hit|hitid)}}">GenBank</a> + <a class=linkheader href="{{genelink(hit|hitid, 'graph')}}">Graphics</a> </div> - + <div class=title> <p class=hittitle>{{hit|firsttitle}}</p> <p class=titleinfo> - <span class=b>Sequence ID:</span> <a href="http://www.ncbi.nlm.nih.gov/nucleotide/{{hit|hitid}}?report=genbank&log$=nuclalign&blast_rank=1">{{hit|seqid}}</a> + <span class=b>Sequence ID:</span> <a href="{{genelink(hit|hitid)}}">{{hit|seqid}}</a> <span class=b>Length:</span> {{hit.Hit_len}} <span class=b>Number of Matches:</span> {{hit.Hit_hsps.Hsp|length}} </p> @@ -473,7 +495,7 @@ <div class=title> <p class=hittitle>{{title.title}}</p> <p class=titleinfo> - <span class=b>Sequence ID:</span> {{title.id}} + <span class=b>Sequence ID:</span> <a href="{{genelink(title.hitid)}}">{{title.id}}</a> </p> </div> {% endfor %} @@ -482,7 +504,11 @@ {% for hsp in hit.Hit_hsps.Hsp %} <div class=hotspot> - <p class=range>Range {{hsp.Hsp_num}}: {{hsp['Hsp_hit-from']}} to {{hsp['Hsp_hit-to']}} <a href="http://www.ncbi.nlm.nih.gov/nucleotide/{{hit|hitid}}?report=genbank&log$=nuclalign&blast_rank=1&from={{hsp['Hsp_hit-from']}}&to={{hsp['Hsp_hit-to']}}">GenBank</a></p> + <p class=range> + <span class=range>Range {{hsp.Hsp_num}}: {{hsp['Hsp_hit-from']}} to {{hsp['Hsp_hit-to']}}</span> + <a class=range href="{{genelink(hit|hitid, 'genbank', hsp)}}">GenBank</a> + <a class=range href="{{genelink(hit|hitid, 'graph', hsp)}}">Graphics</a> + </p> <table class=hotspotstable> <tr>
--- a/visualise.py Mon May 12 13:55:04 2014 +0200 +++ b/visualise.py Mon May 12 17:12:24 2014 +0200 @@ -7,23 +7,22 @@ import math import warnings from itertools import repeat +import argparse from lxml import objectify import jinja2 -blast = objectify.parse('blast xml example1.xml').getroot() -loader = jinja2.FileSystemLoader(searchpath='.') -environment = jinja2.Environment(loader=loader, lstrip_blocks=True, trim_blocks=True, autoescape=True) +_filters = {} def filter(func_or_name): "Decorator to register a function as filter in the current jinja environment" if isinstance(func_or_name, str): def inner(func): - environment.filters[func_or_name] = func + _filters[func_or_name] = func return func return inner else: - environment.filters[func_or_name.__name__] = func_or_name + _filters[func_or_name.__name__] = func_or_name return func_or_name @@ -38,10 +37,6 @@ return 3 return 4 -colors = ['black', 'blue', 'green', 'magenta', 'red'] - -environment.filters['color'] = lambda length: match_colors[color_idx(length)] - @filter def fmt(val, fmt): return format(float(val), fmt) @@ -58,8 +53,9 @@ titles = [] for t in id_titles[1:]: fullid, title = t.split(' ', 1) - id = fullid.split('|', 2)[2] + hitid, id = fullid.split('|', 2)[1:3] titles.append(dict(id = id, + hitid = hitid, fullid = fullid, title = title)) return titles @@ -92,108 +88,158 @@ return 'Minus' raise Exception("frame should be either +1 or -1") - -query_length = int(blast["BlastOutput_query-len"]) - -hits = blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit -# sort hits by longest hotspot first -ordered_hits = sorted(hits, - key=lambda h: max(hsplen(hsp) for hsp in h.Hit_hsps.Hsp), - reverse=True) - -def match_colors(): - """ - An iterator that yields lists of length-color pairs. - """ +def genelink(hit, type='genbank', hsp=None): + if not isinstance(hit, str): + hit = hitid(hit) + link = "http://www.ncbi.nlm.nih.gov/nucleotide/{}?report={}&log$=nuclalign".format(hit, type) + if hsp != None: + link += "&from={}&to={}".format(hsp['Hsp_hit-from'], hsp['Hsp_hit-to']) + return jinja2.Markup(link) - percent_multiplier = 100 / query_length - - for hit in hits: - # sort hotspots from short to long, so we can overwrite index colors of - # short matches with those of long ones. - hotspots = sorted(hit.Hit_hsps.Hsp, key=lambda hsp: hsplen(hsp)) - table = bytearray([255]) * query_length - for hsp in hotspots: - frm = hsp['Hsp_query-from'] - 1 - to = int(hsp['Hsp_query-to']) - table[frm:to] = repeat(color_idx(hsplen(hsp)), to - frm) - matches = [] - last = table[0] - count = 0 - for i in range(query_length): - if table[i] == last: - count += 1 - continue - matches.append((count * percent_multiplier, colors[last] if last != 255 else 'none')) - last = table[i] - count = 1 - matches.append((count * percent_multiplier, colors[last] if last != 255 else 'none')) - - yield dict(colors=matches, link="#hit"+hit.Hit_num.text, defline=firsttitle(hit)) -def queryscale(): - max_labels = 10 - skip = math.ceil(query_length / max_labels) - percent_multiplier = 100 / query_length - for i in range(1, query_length+1): - if i % skip == 0: - yield dict(label = i, width = skip * percent_multiplier) - if query_length % skip != 0: - yield dict(label = query_length, width = (query_length % skip) * percent_multiplier) +class BlastVisualize: + + colors = ('black', 'blue', 'green', 'magenta', 'red') + + max_scale_labels = 10 + + templatename = 'visualise.html.jinja' + + def __init__(self, input): + self.input = input + + self.blast = objectify.parse(self.input).getroot() + self.loader = jinja2.FileSystemLoader(searchpath='.') + self.environment = jinja2.Environment(loader=self.loader, + lstrip_blocks=True, trim_blocks=True, autoescape=True) + + self.environment.filters['color'] = lambda length: match_colors[color_idx(length)] + + for name, filter in _filters.items(): + self.environment.filters[name] = filter + + self.query_length = int(self.blast["BlastOutput_query-len"]) + self.hits = self.blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit + # sort hits by longest hotspot first + self.ordered_hits = sorted(self.hits, + key=lambda h: max(hsplen(hsp) for hsp in h.Hit_hsps.Hsp), + reverse=True) + + def render(self, output): + template = self.environment.get_template(self.templatename) + + params = (('Query ID', self.blast["BlastOutput_query-ID"]), + ('Query definition', self.blast["BlastOutput_query-def"]), + ('Query length', self.blast["BlastOutput_query-len"]), + ('Program', self.blast.BlastOutput_version), + ('Database', self.blast.BlastOutput_db), + ) + + if len(self.blast.BlastOutput_iterations.Iteration) > 1: + warnings.warn("Multiple 'Iteration' elements found, showing only the first") + + output.write(template.render(blast=self.blast, + length=self.query_length, + hits=self.blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit, + colors=self.colors, + match_colors=self.match_colors(), + queryscale=self.queryscale(), + hit_info=self.hit_info(), + genelink=genelink, + params=params)) + + + def match_colors(self): + """ + An iterator that yields lists of length-color pairs. + """ + + percent_multiplier = 100 / self.query_length + + for hit in self.hits: + # sort hotspots from short to long, so we can overwrite index colors of + # short matches with those of long ones. + hotspots = sorted(hit.Hit_hsps.Hsp, key=lambda hsp: hsplen(hsp)) + table = bytearray([255]) * self.query_length + for hsp in hotspots: + frm = hsp['Hsp_query-from'] - 1 + to = int(hsp['Hsp_query-to']) + table[frm:to] = repeat(color_idx(hsplen(hsp)), to - frm) + + matches = [] + last = table[0] + count = 0 + for i in range(self.query_length): + if table[i] == last: + count += 1 + continue + matches.append((count * percent_multiplier, self.colors[last] if last != 255 else 'none')) + last = table[i] + count = 1 + matches.append((count * percent_multiplier, self.colors[last] if last != 255 else 'none')) + + yield dict(colors=matches, link="#hit"+hit.Hit_num.text, defline=firsttitle(hit)) -def hit_info(): + def queryscale(self): + skip = math.ceil(self.query_length / self.max_scale_labels) + percent_multiplier = 100 / self.query_length + for i in range(1, self.query_length+1): + if i % skip == 0: + yield dict(label = i, width = skip * percent_multiplier) + if self.query_length % skip != 0: + yield dict(label = self.query_length, width = (self.query_length % skip) * percent_multiplier) - for hit in ordered_hits: - hsps = hit.Hit_hsps.Hsp + + def hit_info(self): + + for hit in self.ordered_hits: + hsps = hit.Hit_hsps.Hsp - cover = [False] * query_length - for hsp in hsps: - cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, hsplen(hsp)) - cover_count = cover.count(True) - - def hsp_val(path): - return (hsp[path] for hsp in hsps) - - yield dict(hit = hit, - title = firsttitle(hit), - link_id = hit.Hit_num, - maxscore = "{:.1f}".format(float(max(hsp_val('Hsp_bit-score')))), - totalscore = "{:.1f}".format(float(sum(hsp_val('Hsp_bit-score')))), - cover = "{:.0%}".format(cover_count / query_length), - e_value = "{:.4g}".format(float(min(hsp_val('Hsp_evalue')))), - # FIXME: is this the correct formula vv? - ident = "{:.0%}".format(float(min(hsp.Hsp_identity / hsplen(hsp) for hsp in hsps))), - accession = hit.Hit_accession) + cover = [False] * self.query_length + for hsp in hsps: + cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, hsplen(hsp)) + cover_count = cover.count(True) + + def hsp_val(path): + return (float(hsp[path]) for hsp in hsps) + + yield dict(hit = hit, + title = firsttitle(hit), + link_id = hit.Hit_num, + maxscore = "{:.1f}".format(max(hsp_val('Hsp_bit-score'))), + totalscore = "{:.1f}".format(sum(hsp_val('Hsp_bit-score'))), + cover = "{:.0%}".format(cover_count / self.query_length), + e_value = "{:.4g}".format(min(hsp_val('Hsp_evalue'))), + # FIXME: is this the correct formula vv? + ident = "{:.0%}".format(float(min(hsp.Hsp_identity / hsplen(hsp) for hsp in hsps))), + accession = hit.Hit_accession) def main(): - template = environment.get_template('visualise.html.jinja') - params = (('Query ID', blast["BlastOutput_query-ID"]), - ('Query definition', blast["BlastOutput_query-def"]), - ('Query length', blast["BlastOutput_query-len"]), - ('Program', blast.BlastOutput_version), - ('Database', blast.BlastOutput_db), - ) - - if len(blast.BlastOutput_iterations.Iteration) > 1: - warnings.warn("Multiple 'Iteration' elements found, showing only the first") + parser = argparse.ArgumentParser(description="Convert a BLAST XML result into a nicely readable html page", + usage="{} [-i] INPUT [-o OUTPUT]".format(sys.argv[0])) + input_group = parser.add_mutually_exclusive_group(required=True) + input_group.add_argument('positional_arg', metavar='INPUT', nargs='?', type=argparse.FileType(mode='r'), + help='The input Blast XML file, same as -i/--input') + input_group.add_argument('-i', '--input', type=argparse.FileType(mode='r'), + help='The input Blast XML file') + parser.add_argument('-o', '--output', type=argparse.FileType(mode='w'), default=sys.stdout, + help='The output html file') - sys.stdout.write(template.render(blast=blast, - length=query_length, - hits=blast.BlastOutput_iterations.Iteration.Iteration_hits.Hit, - colors=colors, - match_colors=match_colors(), - queryscale=queryscale(), - hit_info=hit_info(), - params=params)) + args = parser.parse_args() + if args.input == None: + args.input = args.positional_arg + if args.input == None: + parser.error('no input specified') -main() + b = BlastVisualize(args.input) + b.render(args.output) + -# http://www.ncbi.nlm.nih.gov/nucleotide/557804451?report=genbank&log$=nuclalign&blast_rank=1&RID=PHWP1JNZ014 -# http://www.ncbi.nlm.nih.gov/nuccore/557804451?report=graph&rid=PHWP1JNZ014[557804451]&tracks=[key:sequence_track,name:Sequence,display_name:Sequence,id:STD1,category:Sequence,annots:Sequence,ShowLabel:true][key:gene_model_track,CDSProductFeats:false][key:alignment_track,name:other%20alignments,annots:NG%20Alignments%7CRefseq%20Alignments%7CGnomon%20Alignments%7CUnnamed,shown:false]&v=752:2685&appname=ncbiblast&link_loc=fromSubj +if __name__ == '__main__': + main() -# http://www.ncbi.nlm.nih.gov/nucleotide/557804451?report=genbank&log$=nucltop&blast_rank=1&RID=PHWP1JNZ014