Changeset 269

Timestamp:

01/19/08 02:51:20 (10 months ago)

Author:

t

Message:

interface to build PWM from list of input sites

Files:

• trunk/cartwheel-server/lib/cartwheel/website/IUPACMotif.py (modified) (1 diff)
• trunk/cartwheel-server/lib/cartwheel/website/NamedMotif.py (modified) (1 diff)
• trunk/cartwheel-server/lib/cartwheel/website/PWMMotif.py (modified) (2 diffs)
• trunk/cartwheel-server/tests/functional-tests/test-with-user/test-web/test-folder-motifs.py (modified) (1 diff)
• trunk/cartwheel-server/website/canal/motif/__init__.py (modified) (1 diff)
• trunk/cartwheel-server/website/canal/motif/analyze.ptl (modified) (1 diff)
• trunk/cartwheel-server/website/canal/motif/pages.ptl (modified) (6 diffs)

trunk/cartwheel-server/lib/cartwheel/website/IUPACMotif.py

@@ -16 +16 @@
-        return "motif: %s; match %d/%d by default" % \
-               (self.motif, len(self.motif) - self.mismatches, len(self.motif))
+
+    def simple_key(self):
+        return self.motif

trunk/cartwheel-server/lib/cartwheel/website/NamedMotif.py

@@ -5 +5 @@
-    mymembers = ('name', 'motif_type', 'folder_id', 'visible')
-
-
+         # actually runs the search
-        raise NotImplementedError
-
-
+        # info box output
-        raise NotImplementedError
+
+    def simple_key(self):               # key for display tagging
+        raise NotImplementedError

trunk/cartwheel-server/lib/cartwheel/website/PWMMotif.py

@@ +1 @@
+"""
+A PWM motif created from a list of sites.
+"""
-from cartwheel.website.NamedMotif import NamedMotif
-
@@ -4 +7 @@
-    table = 'pwm_motifs'
-    mymembers = ('source_motifs', 'default_threshold')
+
+    def search(self, sequence):
+        """
+        sequence is FolderSequence.
+        """
+        import motility
+
+        sites = self.source_motifs.split("\n")
+        pwm = motility.make_pwm(sites)
+
+        return pwm.find(sequence.seq, self.default_threshold)
+
+    def extra_information(self):
+        import motility
+        
+        sites = self.source_motifs.split("\n")
+        consensus = motility.make_iupac_motif(sites)
+        
+        return "PWM consensus: %s; default threshold %f" % (consensus,
+                                                       self.default_threshold)
+    def simple_key(self):
+        import motility
+        
+        sites = self.source_motifs.split("\n")
+        consensus = motility.make_iupac_motif(sites)
+        return consensus

trunk/cartwheel-server/tests/functional-tests/test-with-user/test-web/test-folder-motifs.py

@@ -167 +167 @@
-    find('18 matches, sorted by position.')
-    back()
+
+    ## end
+    back()
+    follow('Return to folder menu')
+
+def test_add_pwm_motif():
+    show()
+    follow("manage motifs")
+    follow("build_pwm_from_list")
+
+    fv('1', 'name', 'test_pwm_gata')
+    fv('1', 'aligned', 'AGATAG\nTGATAG\nTGATAA')
+
+    submit()
+
+    show()
+
+    find("Matrix motif 'test_pwm_gata' created.")
+    find("lowest recommended threshold: 10.58")
+    find("prob/base with this score: 7.324219e-04")
+
+    follow("Return to motif list")
+
+    ## end
+    follow('Return to folder menu')

trunk/cartwheel-server/website/canal/motif/__init__.py

@@ -5 +5 @@
-import cartwheel.website
-from canal.motif.analyze import AnalyzeSequenceHandler
-
+, build_pwm_from_list
-
-
+
+
-
-def edit(request):

trunk/cartwheel-server/website/canal/motif/analyze.ptl

@@ -26 +26 @@
-        for motif in motifs:
-            motif = self.manager.load(motif.id)
+            key = motif.simple_key()
+            
-            """
-            %s: %d matches <a href='matches_detail?motif_id=%d'
-                              id='detail-%s'>detail</a><br>
-            """ % (motif.name, len(motif.search(self.seq)), motif.id,
-motif.motif
+key
-            
-    template matches_detail(self, request):

trunk/cartwheel-server/website/canal/motif/pages.ptl

@@ -2 +2 @@
-from canal.templates import header, footer
-from cartwheel.website.IUPACMotif import IUPACMotif
+from cartwheel.website.PWMMotif import PWMMotif
-from canal.group.utils import make_sequence_list_widget
-
@@ -7 +8 @@
-import motility
-
-
+
+
-
-template _q_index(request):
@@ -38 +40 @@
-              <td align='right'>
-                <font size='-1'>
-
-
+ id='delete-%s'
+ id='edit-%s'
-                </font>
-              </td>
@@ -46 +48 @@
-                   nice_motif_types[motif.motif_type],
-                   motif.extra_information(),
-
+
+
-
-        """
@@ -58 +61 @@
-    """
-    <hr>
-
-
+
+
+
+
-    <p>
-    """
@@ -155 +160 @@
-
-    return d
+
+def build_pwm_from_list(request):
+    if not request.form:
+        return """\
+<form method="POST">
+<h3>Build a matrix motif (PWM) from a list of known sites</h3>
+Name: <input type='text' name='name'>
+<p>
+Aligned sequences:<br>
+<textarea name='aligned' rows='5' cols='30'></textarea>
+<p>
+<input type='submit' value='build PWM'>
+</form>
+<hr>
+<b>Instructions:</b>
+<p>
+To build a matrix motif (a.k.a. Position-Weight Matrix, or PWM), 
+enter a list of ungapped, aligned sequences.  For example, here is a list
+of GATA-type motifs:
+<pre>
+AGATAG
+TGATAG
+TGATAA
+</pre>
+
+After submitting this list, a matrix motif will be built and you can
+then define a threshold for searches.
+"""
+
+    errors = []
+
+    name = request.form['name'].strip()
+
+    if not name:
+        errors.append("you must enter a name!")
+
+    sequences = request.form['aligned'].split('\n')
+    sequences = [ s.strip().upper() for s in sequences ]
+    sequences = [ s for s in sequences if len(s) ]
+
+    if len(sequences) <= 1:
+        errors.append('you must enter at least two sequences')
+
+    lens = [ len(s) for s in sequences ]
+    if max(lens) != min(lens):
+        errors.append('all sequences must be the same length')
+
+    for s in sequences:
+        if '-' in s:
+            errors.append('one or more of your sequences contains a gap!')
+            break
+
+    # check for only ATCG
+    d = {}
+    for s in sequences:
+        for ch in s:
+            d[ch] = 1
+
+    for ch in 'A', 'T', 'C', 'G', '-':  # already covered gap error, above
+        if ch in d:
+            del d[ch]
+
+    if len(d):
+        errors.append("there is at least one non-DNA character in your sequences; here's a list: %s" % (", ".join(d.keys())))
+
+    if errors:
+        return "ERROR:<p><ul><li> %s</ul>" % ("<li>".join(errors,))
+
+    ### ok, validation good so far.  build the PWM & display...
+    
+    return display_pwm_from_sites(request, name, sequences)
+
+template display_pwm_from_sites(request, name, sequences):
+    header("Motif '%s' created" % (name,))
+    
+    "<h2>Matrix motif '%s' created.</h2>" % (name,)
+    
+    pwm = motility.make_pwm(sequences)
+
+    x = []
+
+    # build index header
+    xx = []
+    for i in range(len(pwm)):
+        xx.append("<td align='center'><i><font size='-1'>%d</font></i></td>" % (i + 1,))
+    x.append("<tr><td></td>" + "".join(xx) + "</tr>")
+
+    dna = 'A', 'C', 'G', 'T'
+    maxes = [ max(pwm.matrix[j]) for j in range(len(pwm)) ]
+
+    for i in range(4):
+        xx = ["<td><font size='-1'><i>%s</i></font></td>" % (dna[i],)]
+
+        for j in range(len(pwm)):
+            val = pwm.matrix[j][i]
+            if val == maxes[j]:
+                xx.append("<td align='center'><b>%.2f</b></td>" % (val,))
+            elif val == 0:
+                xx.append("<td align='center'>-</td>")
+            else:
+                xx.append("<td align='center'>%.2f</td>" % (val,))
+
+        x.append("<tr>" + "".join(xx) + "</tr>")
+
+    # display table
+    "<h3>Matrix:</h3>"
+    "<table border='1'>" + "\n".join(x) + "</table>"
+
+    ###
+
+    min_score = pwm.min_score()
+    max_score = pwm.max_score()
+
+    site_scores = [ (pwm.calc_score(s),s) for s in sequences ]
+    site_scores.sort()
+    site_scores.reverse()
+
+    x = []
+    for score, site in site_scores:
+        x.append('<tr><td>%s:</td><td>%.2f</td>' % (site, score,))
+
+
+    ###
+        
+    "<p>"
+    """
+    <table border='1'>
+    <tr><th>Input site</th><th>PWM score</th></tr>
+    """
+    
+    "".join(x)
+    
+    "</table>"
+
+    ###
+
+    lowest_score = site_scores[-1][0]
+
+    """
+    <p>
+    lowest recommended threshold: %.2f;
+    prob/base with this score: %e
+    """ % (lowest_score, pwm.weight_sites_over(lowest_score))
+
+    ###
+
+    manager = cartwheel.website.get_object_manager()
+
+    folder_id = request.session.get_current_folder().id
+
+    manager.create(PWMMotif, name=name, source_motifs="\n".join(sequences),
+                   motif_type='PWM',
+                   default_threshold=lowest_score,
+                   folder_id=folder_id)
+
+    manager.commit()
+
+    """
+    <p>
+    <a href='./'>Return to motif list</a>
+    """
+
+    footer()