Add chromosomal positions to getTranscriptsAndInfo webservice (#86)

git-svn-id: https://humgenprojects.lumc.nl/svn/mutalyzer/trunk@476 eb6bd6ab-9ccd-42b9-aceb-e2899b4a52f1

extras/soap-tools/getTranscriptsAndInfo.py

@@ -40,15 +40,18 @@ def main(genomic_reference, gene=None):
   Locus tag: %s
   Link method: %s
   Translation:
-    Start: %s (c), %s (g)
-    End: %s (c), %s (g)
+    Start: %s (c), %s (g), %s (chrom)
+    End: %s (c), %s (g), %s (chrom)
     Sortable end: %s
   CDS:
-    Start: %s (c), %s (g)
-    End: %s (c), %s (g)""" % \
+    Start: %s (c), %s (g), %s (chrom)
+    End: %s (c), %s (g), %s (chrom)""" % \
             (t.name, t.id, t.product, t.locusTag, t.linkMethod, t.cTransStart,
-             t.gTransStart, t.cTransEnd, t.gTransEnd, t.sortableTransEnd,
-             t.cCDSStart, t.gCDSStart, t.cCDSStop, t.gCDSStop)
+             t.gTransStart, t.chromTransStart if 'chromTransStart' in t else '-',
+             t.cTransEnd, t.gTransEnd, t.chromTransEnd if 'chromTransEnd' in t else '-',
+             t.sortableTransEnd, t.cCDSStart, t.gCDSStart,
+             t.chromCDSStart if 'chromCDSStart' in t else '-', t.cCDSStop, t.gCDSStop,
+             t.chromCDSStop if 'chromCDSStop' in t else '-')
 
             if 'proteinTranscript' in t:
                 print """  Protein:
@@ -61,8 +64,10 @@ def main(genomic_reference, gene=None):
             if 'exons' in t:
                 print '  Exons:'
                 for e in t.exons.ExonInfo:
-                    print '    %s - %s (c), %s - %s (g)' % \
-                          (e.cStart, e.cStop, e.gStart, e.gStop)
+                    print '    %s - %s (c), %s - %s (g), %s - %s (chrom)' % \
+                          (e.cStart, e.cStop, e.gStart, e.gStop,
+                           e.chromStart if 'chromStart' in e else '-',
+                           e.chromStop if 'chromStop' in e else '-')
 
 
 if __name__ == '__main__':

mutalyzer/GenRecord.py

@@ -351,6 +351,8 @@ class Record(object) :
         @return: chromosomal g. position
         @rtype: integer
         """
+        if not self.chromOffset:
+            return None
 
         if self.orientation == 1 :
             return self.chromOffset + i - 1

mutalyzer/models.py

@@ -147,8 +147,10 @@ class ExonInfo(ClassModel):
 
     cStart = Mandatory.String
     gStart = Mandatory.Integer
+    chromStart = Integer
     cStop = Mandatory.String
     gStop = Mandatory.Integer
+    chromStop = Integer
 #ExonInfo
 
 
@@ -168,9 +170,11 @@ class TranscriptInfo(ClassModel):
     """
     Used in return type of SOAP method getTranscriptsAndInfo.
 
-    @todo: Decide on 'stop' versus 'end'. Web interface uses 'stop' for
-           both trans and CDS. Ivar asked for 'end'. Internally, we have
-           trans 'end' and CDS 'stop'.
+    @todo: Decide on 'stop' versus 'end'. Web interface uses 'stop' for both
+        trans and CDS. Ivar asked for 'end'. Internally, we have trans 'end'
+        and CDS 'stop'.
+    @todo: We should really also provide the chromosome (or its accession
+        number) next to the chromosomal positions, if available.
     """
     __namespace__ = SOAP_NAMESPACE
 
@@ -180,14 +184,18 @@ class TranscriptInfo(ClassModel):
 
     cTransStart = Mandatory.String
     gTransStart = Mandatory.Integer
+    chromTransStart = Integer
     cTransEnd = Mandatory.String
     gTransEnd = Mandatory.Integer
+    chromTransEnd = Integer
     sortableTransEnd = Mandatory.Integer
 
     cCDSStart = Mandatory.String
     gCDSStart = Mandatory.Integer
+    chromCDSStart = Integer
     cCDSStop = Mandatory.String
     gCDSStop = Mandatory.Integer
+    chromCDSStop = Integer
 
     locusTag = Mandatory.String
     linkMethod = Mandatory.String

mutalyzer/parsers/genbank.py

@@ -501,7 +501,11 @@ class GBparser():
         geneDict = {}
 
         accInfo = biorecord.annotations['accessions']
-        if len(accInfo) >= 3 and accInfo[1] == "REGION:" :
+        if len(accInfo) >= 3 and accInfo[1] == "REGION:":
+            # Todo: This information is present in the genbank file if it is a
+            #     UD sliced from a chromosome. We can get the same information
+            #     for NM references from our mapping database and that way
+            #     also provide chromosomal variant descriptions for those.
             region = accInfo[2]
             if "complement" in region :
                 record.orientation = -1

mutalyzer/webservice.py

@@ -787,20 +787,26 @@ class MutalyzerService(DefinitionBase):
                  - product
                  - cTransStart
                  - gTransStart
+                 - chromTransStart
                  - cTransEnd
                  - gTransEnd
+                 - chromTransEnd
                  - sortableTransEnd
                  - cCDSStart
                  - gCDSStart
+                 - chromCDSStart
                  - cCDSStop
                  - gCDSStop
+                 - chromCDSStop
                  - locusTag
                  - linkMethod
                  - exons: Array of ExonInfo objects with fields:
                           - cStart
                           - gStart
+                          - chromStart
                           - cStop
                           - gStop
+                          - chromStop
                  - proteinTranscript: ProteinTranscript object with fields:
                                       - name
                                       - id
@@ -847,8 +853,10 @@ class MutalyzerService(DefinitionBase):
                     exon = ExonInfo()
                     exon.gStart = transcript.CM.getSpliceSite(i)
                     exon.cStart = transcript.CM.g2c(exon.gStart)
+                    exon.chromStart = GenRecordInstance.record.toChromPos(exon.gStart)
                     exon.gStop = transcript.CM.getSpliceSite(i + 1)
                     exon.cStop = transcript.CM.g2c(exon.gStop)
+                    exon.chromStop = GenRecordInstance.record.toChromPos(exon.gStop)
                     t.exons.append(exon)
 
                 # Beware that CM.info() gives a made-up value for trans_end,
@@ -861,6 +869,7 @@ class MutalyzerService(DefinitionBase):
 
                 t.cTransEnd = str(t.exons[-1].cStop)
                 t.gTransEnd = t.exons[-1].gStop
+                t.chromTransEnd = GenRecordInstance.record.toChromPos(t.gTransEnd)
                 t.sortableTransEnd = sortable_trans_end
 
                 # Todo: If we have no CDS info, CM.info() gives trans_end as
@@ -873,10 +882,13 @@ class MutalyzerService(DefinitionBase):
                 t.product = transcript.transcriptProduct
                 t.cTransStart = str(trans_start)
                 t.gTransStart = transcript.CM.x2g(trans_start, 0)
+                t.chromTransStart = GenRecordInstance.record.toChromPos(t.gTransStart)
                 t.cCDSStart = str(cds_start)
                 t.gCDSStart = transcript.CM.x2g(cds_start, 0)
+                t.chromCDSStart = GenRecordInstance.record.toChromPos(t.gCDSStart)
                 t.cCDSStop = str(cds_stop)
                 t.gCDSStop = transcript.CM.x2g(cds_stop, 0)
+                t.chromCDSStop = GenRecordInstance.record.toChromPos(t.gCDSStop)
                 t.locusTag = transcript.locusTag
                 t.linkMethod = transcript.linkMethod
 
@@ -948,6 +960,10 @@ class MutalyzerService(DefinitionBase):
     def sliceChromosome(self, chromAccNo, start, end, orientation) :
         """
         Todo: documentation, error handling, argument checking, tests.
+
+        @arg orientation: Orientation of the slice. 1 for forward, 2 for
+            reverse complement.
+        @type orientation: integer
         """
         O = Output(__file__)
         D = Db.Cache()

tests/test_webservice.py

@@ -59,8 +59,8 @@ class TestWebservice():
         @todo: Start the standalone server and stop it in self.tearDown
         instead of depending on some running instance at a fixed address.
         """
-        self.client = Client(WSDL_URL, cache=None)
-        #self.client.options.cache.setduration(seconds=120)
+        self.client = Client(WSDL_URL) #, cache=None)
+        self.client.options.cache.setduration(seconds=120)
 
     def test_checksyntax_valid(self):
         """
@@ -400,3 +400,37 @@ class TestWebservice():
         assert_equal(r.sourceVersion, '1')
         assert_equal(r.sourceGi, '256574794')
         assert_equal(r.molecule, 'g')
+
+    @skip
+    def test_gettranscriptsandinfo_slice(self):
+        """
+        Running getTranscriptsAndInfo on a chromosomal slice should include
+        chromosomal positions.
+
+        slice: 48284000 - 48259456 (COL1A1 with 5001 and 2001 borders)
+        translation start: 48284000 - 5001 + 1 = 48279000
+        translation end: 48259456 + 2001 = 48261457
+
+        Todo: We cannot use UD references in unit tests, unless we implement
+           a way to create them inside the unit test.
+        """
+        r = self.client.service.getTranscriptsAndInfo('UD_129646580028')
+        assert_equal(type(r.TranscriptInfo), list)
+        names = [t.name for t in r.TranscriptInfo]
+        assert 'COL1A1_v001' in names
+        for t in r.TranscriptInfo:
+            if t.name != 'COL1A1_v001':
+                continue
+            assert_equal(t.cTransStart, '-126')
+            assert_equal(t.gTransStart, 5001)
+            assert_equal(t.chromTransStart, 48279000)
+            assert_equal(t.cTransEnd, '*1406')
+            assert_equal(t.gTransEnd, 22544)
+            assert_equal(t.chromTransEnd, 48261457)
+            assert_equal(t.sortableTransEnd, 5801)
+            assert_equal(t.cCDSStart, '1')
+            assert_equal(t.gCDSStart, 5127)
+            assert_equal(t.chromCDSStart, 48278874)
+            assert_equal(t.cCDSStop, '4395')
+            assert_equal(t.gCDSStop, 21138)
+            assert_equal(t.chromCDSStop, 48262863)