Mercurial > repos > jankanis > blast2html

changeset 19:67ddcb807b7d

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
make it work with multiple queries
author Jan Kanis <jan.code@jankanis.nl>
date Tue, 13 May 2014 18:06:36 +0200
parents 4434ffab721a
children 53cd304c5f26
files blast_html.html.jinja blast_html.py
diffstat 2 files changed, 250 insertions(+), 221 deletions(-) [+]
line wrap: on
line diff
--- a/blast_html.html.jinja	Tue May 13 15:26:20 2014 +0200
+++ b/blast_html.html.jinja	Tue May 13 18:06:36 2014 +0200
@@ -2,6 +2,7 @@
 <html>
   <head>
     <meta charset="UTF-8">
+    <meta name=generator content="blast_html; see ...">
     
     <title>Blast output</title>
     
@@ -97,12 +98,12 @@
       margin: auto;
       }
 
-      .centered, #defline, div.legend, div.tablewrapper {
+      .centered, .defline, div.legend, div.tablewrapper {
       margin-left: auto;
       margin-right: auto;
       }
 
-      #defline {
+      .defline {
       background-color: white;
       border: 1px solid black;
       margin: .5em auto;
@@ -338,197 +339,213 @@
   
   <body>
     <div id=content>
-      <h1>Nucleotide Sequence ({{length}} letters)</h1>
 
-      <section class=header>
-
-        <table class=headerdata>
-          {% for param, value in params %}
-            <tr><td class=param>{{param}}:</td><td>{{value}}</td></tr>
-          {% endfor %}
-        </table>
-        
-      </section>
-      
-      {% if not (blast.BlastOutput_iterations.findall('Iteration') and
-                 blast.BlastOutput_iterations.Iteration.Iteration_hits.findall('Hit')) %}
+      {% if not blast.BlastOutput_iterations.findall('Iteration') %}
       <section class=nodata>
-        <h2>No Results</h2>
-        <div class=grey>
-          No Matches
-        </div>
-      </section>
-      {% else %}
-
-
-      
-      <section class=graphics>
-        <h2>Graphic Summary</h2>
-
+        <h1>No data</h1>
         <div class=grey>
-          <h3 class=centered>Distribution of {{hits|length}} Blast Hits on the Query Sequence</h3>
-          
-          <div id=defline>Mouse-over to show defline and scores, click to show alignments</div>
-
-          <div class=graphic>
-            <h4 class=darkHeader>Color key for alignment scores</h4>
-            <div class=legend><div class=graphicrow>
-                <div class=graphicitem style="background-color: {{colors[0]}}">&lt;40</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>
-
-            <div class=tablewrapper>
-
-              <div class=scale>
-                <div>query:</div>
-                <div class=graphicrow>
-                  {% for s in queryscale %}
-                  <div class=graphicitem style="width: {{s.width}}%">
-                    <div>{{s.label}}</div>
-                  </div>
-                  {% endfor %}
-                </div>
-                <div style="clear: left"></div>
-              </div>
-              
-              {% for line in match_colors %}
-              <a class=matchresult
-                 href="{{line.link}}"
-                 onmouseover='document.getElementById("defline").innerHTML="{{line.defline|js_string_escape}}"'
-                 onmouseout='document.getElementById("defline").innerHTML="Mouse-over to show defline and scores, click to show alignments"'
-                 title="{{line.defline}}">
-                <div class="matchrow graphicrow">
-                  {% for match in line.colors %}
-                  <div class="matchitem graphicitem"
-                       style="background-color: {{match[1]}}; width: {{match[0]}}%"></div>
-                  {% endfor %}
-                </div>
-              </a>
-              
-              {% endfor %}
-            </div>
-          </div>
+          No matches
         </div>
       </section>
 
-
+      {% else %}
+      {% for result in blast.BlastOutput_iterations.Iteration %}
+      
+      <section class=match>
       
-      <section class=descriptions>
-        <h2>Descriptions</h2>
+        <h1>Nucleotide Sequence ({{result|len}} letters)</h1>
+
+        <section class=header>
+
+          <table class=headerdata>
+            {% for param, value in params %}
+              <tr><td class=param>{{param}}:</td><td>{{value}}</td></tr>
+            {% endfor %}
+          </table>
+
+        </section>
+
+        {% set hits = result|hits %}
+        {% if not hits %}
+        <section>
+          <h2>No Hits</h2>
+          <div class=grey>
+            This query did not match anywhere
+          </div>
+        </section>
+        {% else %}
+
+        <section class=graphics>
+          <h2>Graphic Summary</h2>
+
+          <div class=grey>
+            <h3 class=centered>Distribution of {{result|length}} Blast Hits on the Query Sequence</h3>
+
+            <div class=defline id=defline{{result['Iteration_iter-num']}}>
+              Mouse-over to show defline and scores, click to show alignments
+            </div>
 
-        <div class=grey><div class=white>
-            <h4 class=darkHeader>Sequences producing significant alignments:</h4>
+            <div class=graphic>
+              <h4 class=darkHeader>Color key for alignment scores</h4>
+              <div class=legend><div class=graphicrow>
+                  <div class=graphicitem style="background-color: {{colors[0]}}">&lt;40</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>
+
+              <div class=tablewrapper>
+
+                <div class=scale>
+                  <div>query:</div>
+                  <div class=graphicrow>
+                    {% for s in result|queryscale %}
+                    <div class=graphicitem style="width: {{s.width}}%">
+                      <div>{{s.label}}</div>
+                    </div>
+                    {% endfor %}
+                  </div>
+                  <div style="clear: left"></div>
+                </div>
+
+                {% for line in result|match_colors %}
+                <a class=matchresult
+                   href="{{line.link}}"
+                   onmouseover='document.getElementById("defline{{result['Iteration_iter-num']}}").innerHTML="{{line.defline|js_string_escape}}"'
+                   onmouseout='document.getElementById("defline{{result['Iteration_iter-num']}}").innerHTML="Mouse-over to show defline and scores, click to show alignments"'
+                   title="{{line.defline}}">
+                  <div class="matchrow graphicrow">
+                    {% for hit in line.colors %}
+                    <div class="matchitem graphicitem"
+                         style="background-color: {{hit[1]}}; width: {{hit[0]}}%"></div>
+                    {% endfor %}
+                  </div>
+                </a>
+
+                {% endfor %}
+              </div>
+            </div>
+          </div>
+        </section>
+
+
+
+        <section class=descriptions>
+          <h2>Descriptions</h2>
+
+          <div class=grey><div class=white>
+              <h4 class=darkHeader>Sequences producing significant alignments:</h4>
 
-            <table class=descriptiontable>
-              <col/><col/><col/><col/><col/><col/><col/>
-              <tr>
-                <th>Description</th>
-                <th>Max score</th>
-                <th>Total score</th>
-                <th>Query cover</th>
-                <th>E value</th>
-                <th>Ident</th>
-                <th>Accession</th>
-              </tr>
-              {% for hit in hit_info %}
-              <tr>
-                <td><div><a href="#hit{{hit.link_id}}"
-                            title="{{hit.title}}"
-                            id="description{{hit.link_id}}">
-                      {{hit.title}}
-                </a></div></td>
-                <td>{{hit.maxscore}}</td>
-                <td>{{hit.totalscore}}</td>
-                <td>{{hit.cover}}</td>
-                <td>{{hit.e_value}}</td>
-                <td>{{hit.ident}}</td>
-                <td><a href="{{genelink(hit.hit|hitid)}}">{{hit.accession}}</a></td>
-              </tr>
-              {% endfor %}
-            </table>
-            
-        </div></div>
-      </section>
+              <table class=descriptiontable>
+                <col/><col/><col/><col/><col/><col/><col/>
+                <tr>
+                  <th>Description</th>
+                  <th>Max score</th>
+                  <th>Total score</th>
+                  <th>Query cover</th>
+                  <th>E value</th>
+                  <th>Ident</th>
+                  <th>Accession</th>
+                </tr>
+                {% for hit in result|hit_info %}
+                <tr>
+                  <td><div><a href="#hit{{hit.link_id}}"
+                              title="{{hit.title}}"
+                              id="description{{hit.link_id}}">
+                        {{hit.title}}
+                  </a></div></td>
+                  <td>{{hit.maxscore}}</td>
+                  <td>{{hit.totalscore}}</td>
+                  <td>{{hit.cover}}</td>
+                  <td>{{hit.e_value}}</td>
+                  <td>{{hit.ident}}</td>
+                  <td><a href="{{genelink(hit.hit|hitid)}}">{{hit.accession}}</a></td>
+                </tr>
+                {% endfor %}
+              </table>
+
+          </div></div>
+        </section>
+
+
+
+        <section class=alignments>
+          <h2>Alignments</h2>
+
+          <div class=grey><div class=white>
+              {% for hit in hits %}
+              <div class=alignment id=hit{{hit.Hit_num}}>
+
+                <div class=linkheader>
+                  <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="{{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>
+                </div>
+
+                {% if hit|othertitles|length %}
+                <a class=showmoretitles onclick="toggle_visibility('moretitles{{hit.Hit_num|js_string_escape}}'); return false;" href=''>
+                  See {{hit|othertitles|length}} more title(s)
+                </a>
+
+                <div class=moretitles id=moretitles{{hit.Hit_num}} style="display: none">
+                  {% for title in hit|othertitles %}
+                  <div class=title>
+                    <p class=hittitle>{{title.title}}</p>
+                    <p class=titleinfo>
+                      <span class=b>Sequence ID:</span> <a href="{{genelink(title.hitid)}}">{{title.id}}</a>
+                    </p>
+                  </div>
+                  {% endfor %}
+                </div>
+                {% endif %}
 
-      <section class=alignments>
-        <h2>Alignments</h2>
-        
-        <div class=grey><div class=white>
-            {% for hit in hits %}
-            <div class=alignment id=hit{{hit.Hit_num}}>
-              
-              <div class=linkheader>
-                <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="{{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>
+                {% for hsp in hit.Hit_hsps.Hsp %}
+                <div class=hotspot>
+                  <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>
+                      <th>Score</th><th>Expect</th><th>Identities</th><th>Gaps</th><th>Strand</th>
+                    </tr>
+                    <tr>
+                      <td>{{hsp['Hsp_bit-score']|fmt('.1f')}} bits({{hsp.Hsp_score}})</td>
+                      <td>{{hsp.Hsp_evalue|fmt('.1f')}}</td>
+                      <td>{{ hsp.Hsp_identity }}/{{ hsp|len }}({{
+                        (hsp.Hsp_identity/hsp|len) |fmt('.0%') }})</td>
+                      <td>{{ hsp.Hsp_gaps }}/{{ hsp|len
+                        }}({{ (hsp.Hsp_gaps / hsp|len) | fmt('.0%') }})</td>
+                      <td>{{ hsp['Hsp_query-frame']|asframe }}/{{ hsp['Hsp_hit-frame']|asframe }}</td>
+                    </tr>
+                  </table>
+
+                  <pre class=alignmentgraphic>{{hsp|alignment_pre}}</pre>
+                </div>
+                {% endfor %}
+
               </div>
 
-              {% if hit|othertitles|length %}
-              <a class=showmoretitles onclick="toggle_visibility('moretitles{{hit.Hit_num|js_string_escape}}'); return false;" href=''>
-                See {{hit|othertitles|length}} more title(s)
-              </a>
-
-              <div class=moretitles id=moretitles{{hit.Hit_num}} style="display: none">
-                {% for title in hit|othertitles %}
-                <div class=title>
-                  <p class=hittitle>{{title.title}}</p>
-                  <p class=titleinfo>
-                    <span class=b>Sequence ID:</span> <a href="{{genelink(title.hitid)}}">{{title.id}}</a>
-                  </p>
-                </div>
-                {% endfor %}
-              </div>
-              {% endif %}
-
-              {% for hsp in hit.Hit_hsps.Hsp %}
-              <div class=hotspot>
-                <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>
+              {% endfor %}
+          </div></div>
+        </section>
+        {% endif %}
 
-                <table class=hotspotstable>
-                  <tr>
-                    <th>Score</th><th>Expect</th><th>Identities</th><th>Gaps</th><th>Strand</th>
-                  </tr>
-                  <tr>
-                    <td>{{hsp['Hsp_bit-score']|fmt('.1f')}} bits({{hsp.Hsp_score}})</td>
-                    <td>{{hsp.Hsp_evalue|fmt('.1f')}}</td>
-                    <td>{{ hsp.Hsp_identity }}/{{ hsp|len }}({{
-                      (hsp.Hsp_identity/hsp|len) |fmt('.0%') }})</td>
-                    <td>{{ hsp.Hsp_gaps }}/{{ hsp|len
-                      }}({{ (hsp.Hsp_gaps / hsp|len) | fmt('.0%') }})</td>
-                    <td>{{ hsp['Hsp_query-frame']|asframe }}/{{ hsp['Hsp_hit-frame']|asframe }}</td>
-                  </tr>
-                </table>
-
-                <pre class=alignmentgraphic>{{hsp|alignment_pre}}</pre>
-              </div>
-              {% endfor %}
-              
-            </div>
-
-            {% endfor %}
-        </div></div>
-      </section>
-
-      {% endif %}
+        {% endfor %}
+        {% endif %}
     </div>
   </body>
 </html>
--- a/blast_html.py	Tue May 13 15:26:20 2014 +0200
+++ b/blast_html.py	Tue May 13 18:06:36 2014 +0200
@@ -19,11 +19,11 @@
     "Decorator to register a function as filter in the current jinja environment"
     if isinstance(func_or_name, str):
         def inner(func):
-            _filters[func_or_name] = func
+            _filters[func_or_name] = func.__name__
             return func
         return inner
     else:
-        _filters[func_or_name.__name__] = func_or_name
+        _filters[func_or_name.__name__] = func_or_name.__name__
         return func_or_name
 
 
@@ -78,8 +78,13 @@
     )
 
 @filter('len')
-def hsplen(node):
-    return int(node['Hsp_align-len'])
+def blastxml_len(node):
+    if node.tag == 'Hsp':
+        return int(node['Hsp_align-len'])
+    elif node.tag == 'Iteration':
+        return int(node['Iteration_query-len'])
+    raise Exception("Unknown XML node type: "+node.tag)
+        
 
 @filter
 def asframe(frame):
@@ -134,6 +139,13 @@
 
     return value
 
+@filter
+def hits(result):
+    # sort hits by longest hotspot first
+    return sorted(result.Iteration_hits.findall('Hit'),
+                  key=lambda h: max(blastxml_len(hsp) for hsp in h.Hit_hsps.Hsp),
+                  reverse=True)
+
 
 
 class BlastVisualize:
@@ -151,15 +163,15 @@
         self.environment = jinja2.Environment(loader=self.loader,
                                               lstrip_blocks=True, trim_blocks=True, autoescape=True)
 
-        self.environment.filters.update(_filters)
-        self.environment.filters['color'] = lambda length: match_colors[color_idx(length)]
+        self._addfilters(self.environment)
+
 
-        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 _addfilters(self, environment):
+        for filtername, funcname in _filters.items():
+            try:
+                environment.filters[filtername] = getattr(self, funcname)
+            except AttributeError:
+                environment.filters[filtername] = globals()[funcname]
 
     def render(self, output):
         template = self.environment.get_template(self.templatename)
@@ -171,41 +183,38 @@
                   ('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,
+                                     iterations=self.blast.BlastOutput_iterations.Iteration,
                                      colors=self.colors,
-                                     match_colors=self.match_colors(),
-                                     queryscale=self.queryscale(),
-                                     hit_info=self.hit_info(),
+                                     # match_colors=self.match_colors(),
+                                     # hit_info=self.hit_info(),
                                      genelink=genelink,
                                      params=params))
-        
 
-    def match_colors(self):
+    @filter
+    def match_colors(self, result):
         """
         An iterator that yields lists of length-color pairs. 
         """
 
-        percent_multiplier = 100 / self.query_length
+        query_length = blastxml_len(result)
+        
+        percent_multiplier = 100 / query_length
 
-        for hit in self.hits:
+        for hit in hits(result):
             # 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
+            hotspots = sorted(hit.Hit_hsps.Hsp, key=lambda hsp: blastxml_len(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)
+                table[frm:to] = repeat(color_idx(blastxml_len(hsp)), to - frm)
 
             matches = []
             last = table[0]
             count = 0
-            for i in range(self.query_length):
+            for i in range(query_length):
                 if table[i] == last:
                     count += 1
                     continue
@@ -216,25 +225,28 @@
 
             yield dict(colors=matches, link="#hit"+hit.Hit_num.text, defline=firsttitle(hit))
 
-
-    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):
+    @filter
+    def queryscale(self, result):
+        query_length = blastxml_len(result)
+        skip = math.ceil(query_length / self.max_scale_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 self.query_length % skip != 0:
-            yield dict(label = self.query_length, width = (self.query_length % skip) * percent_multiplier)
-
+        if query_length % skip != 0:
+            yield dict(label = query_length, width = (query_length % skip) * percent_multiplier)
 
-    def hit_info(self):
+    @filter
+    def hit_info(self, result):
 
-        for hit in self.ordered_hits:
+        query_length = blastxml_len(result)
+
+        for hit in hits(result):
             hsps = hit.Hit_hsps.Hsp
 
-            cover = [False] * self.query_length
+            cover = [False] * query_length
             for hsp in hsps:
-                cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, hsplen(hsp))
+                cover[hsp['Hsp_query-from']-1 : int(hsp['Hsp_query-to'])] = repeat(True, blastxml_len(hsp))
             cover_count = cover.count(True)
 
             def hsp_val(path):
@@ -245,10 +257,10 @@
                        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),
+                       cover = "{:.0%}".format(cover_count / 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))),
+                       ident = "{:.0%}".format(float(min(hsp.Hsp_identity / blastxml_len(hsp) for hsp in hsps))),
                        accession = hit.Hit_accession)
 
 def main():