From 646ce812f052219048762b012d45337b0345492d Mon Sep 17 00:00:00 2001
From: "J.F.J. Laros" <j.f.j.laros@lumc.nl>
Date: Sun, 19 Aug 2012 13:51:35 +0000
Subject: [PATCH] Added description extractor functionality to the name checker
as an experimental service.
describe.py:
- Added an `unknown' type to the possible types of raw variants.
- Disabled the experimental LCSMatrix() code.
git-svn-id: https://humgenprojects.lumc.nl/svn/mutalyzer/trunk@596 eb6bd6ab-9ccd-42b9-aceb-e2899b4a52f1
---
mutalyzer/describe.py | 17 +++++++++++++++--
mutalyzer/templates/check.html | 14 ++++++++++++++
mutalyzer/website.py | 13 ++++++++++++-
3 files changed, 41 insertions(+), 3 deletions(-)
diff --git a/mutalyzer/describe.py b/mutalyzer/describe.py
index 644573a7..1863bb9d 100644
--- a/mutalyzer/describe.py
+++ b/mutalyzer/describe.py
@@ -298,6 +298,8 @@ class RawVar(models.RawVar):
@returns: The HGVS description of the raw variant stored in this class.
@rtype: str
"""
+ if self.type == "unknown":
+ return "?"
if not self.start:
return "="
@@ -317,6 +319,7 @@ class RawVar(models.RawVar):
descr += self.type
if self.inserted:
descr += "%s" % seq3(self.inserted)
+
if self.type == "stop":
return descr + '*'
if self.term:
@@ -336,7 +339,7 @@ class RawVar(models.RawVar):
variant stored in this class.
@rtype: int
"""
- if not self.start : # =
+ if not self.start : # `=' or `?'
return 1
descrLen = 1 # Start position.
@@ -625,6 +628,13 @@ def protein_description(M, s1, s2, s1_start, s1_end, s2_start, s2_end):
@returns: A list of RawVar objects, representing the allele.
@rval: list(RawVar)
"""
+ if s1 == '?' or s2 == '?':
+ return [RawVar(DNA=False, type="unknown")]
+
+ # One of the sequences is missing.
+ if not (s1 and s2):
+ return [RawVar(DNA=False)]
+
# Nothing happened.
if s1 == s2:
return [RawVar(DNA=False)]
@@ -743,7 +753,10 @@ def describe(original, mutated, DNA=True):
s1_end = len(s1) - lcs
s2_end = len(s2) - lcs
- M = LCSMatrix(s1, s2)
+ # TODO: use only the altered part for the rest of the analysis.
+ # the lcp and lcs can be ignored.
+ #M = LCSMatrix(s1, s2)
+ M = []
if not DNA:
return protein_description(M, s1, s2, lcp, s1_end, lcp, s2_end)
diff --git a/mutalyzer/templates/check.html b/mutalyzer/templates/check.html
index 758a9a4d..89da7bb0 100644
--- a/mutalyzer/templates/check.html
+++ b/mutalyzer/templates/check.html
@@ -268,7 +268,21 @@
<br>
<br>
</div> <!-- interactive -->
+ <br>
</div> <!-- reference -->
+ <div>
+ <b>Experimental services:</b><br>
+ <br>
+ Genomic description:
+ <tt>
+ <div tal:replace = "structure extractedDescription/0"></div><br>
+ </tt>
+ <br>
+ Protein description:
+ <tt>
+ <div tal:replace = "structure extractedProtein/0"></div><br>
+ </tt>
+ </div> <!-- experimental -->
</div> <!-- name -->
</div>
</body>
diff --git a/mutalyzer/website.py b/mutalyzer/website.py
index f7d706ce..89ee649d 100644
--- a/mutalyzer/website.py
+++ b/mutalyzer/website.py
@@ -824,6 +824,15 @@ class Check:
chromosome=raw_variants[0], start=min(positions) - 10,
stop=max(positions) + 10, bed_file=urllib.quote(bed_url))
+ extracted = describe.alleleDescription(
+ describe.describe(output.getIndexedOutput("original", 0),
+ output.getIndexedOutput("mutated", 0)))
+
+ extractedProt = describe.alleleDescription(
+ describe.describe(output.getIndexedOutput("oldprotein", 0),
+ output.getIndexedOutput("newprotein", 0, default=""), DNA=False))
+
+
# Todo: Generate the fancy HTML views for the proteins here instead
# of in mutalyzer/variantchecker.py.
args = {
@@ -852,7 +861,9 @@ class Check:
'restrictionSites' : output.getOutput('restrictionSites'),
'legends' : output.getOutput('legends'),
'reference' : reference,
- 'browserLink' : browser_link
+ 'browserLink' : browser_link,
+ 'extractedDescription' : (extracted, urllib.quote(extracted)),
+ 'extractedProtein' : (extractedProt, urllib.quote(extractedProt))
}
output.addMessage(__file__, -1, 'INFO', 'Finished variant %s' % name)
--
GitLab