/*
* need a sourceDbRef if we are to construct dbrefs to the CDS
- * sequence
+ * sequence from the dna contig sequences
*/
DBRefEntry dbref = new DBRefEntry("ENSEMBL", "0", "dna1");
- dna1.getDatasetSequence().setSourceDBRef(dbref);
+ dna1.getDatasetSequence().addDBRef(dbref);
+ org.testng.Assert.assertEquals(dbref, dna1.getPrimaryDBRefs().get(0));
dbref = new DBRefEntry("ENSEMBL", "0", "dna2");
- dna2.getDatasetSequence().setSourceDBRef(dbref);
+ dna2.getDatasetSequence().addDBRef(dbref);
+ org.testng.Assert.assertEquals(dbref, dna2.getPrimaryDBRefs().get(0));
/*
* CDS sequences are 'discovered' from dna-to-protein mappings on the alignment
* dataset (e.g. added from dbrefs by CrossRef.findXrefSequences)
*/
- MapList map = new MapList(new int[] { 4, 6, 10, 12 },
+ MapList mapfordna1 = new MapList(new int[] { 4, 6, 10, 12 },
new int[] { 1, 2 }, 3, 1);
AlignedCodonFrame acf = new AlignedCodonFrame();
- acf.addMap(dna1.getDatasetSequence(), pep1.getDatasetSequence(), map);
+ acf.addMap(dna1.getDatasetSequence(), pep1.getDatasetSequence(),
+ mapfordna1);
dna.addCodonFrame(acf);
- map = new MapList(new int[] { 1, 3, 7, 9, 13, 15 }, new int[] { 1, 3 },
+ MapList mapfordna2 = new MapList(new int[] { 1, 3, 7, 9, 13, 15 },
+ new int[] { 1, 3 },
3, 1);
acf = new AlignedCodonFrame();
- acf.addMap(dna2.getDatasetSequence(), pep2.getDatasetSequence(), map);
+ acf.addMap(dna2.getDatasetSequence(), pep2.getDatasetSequence(),
+ mapfordna2);
dna.addCodonFrame(acf);
/*
+ * In this case, mappings originally came from matching Uniprot accessions - so need an xref on dna involving those regions. These are normally constructed from CDS annotation
+ */
+ DBRefEntry dna1xref = new DBRefEntry("UNIPROT", "ENSEMBL", "pep1",
+ new Mapping(mapfordna1));
+ dna1.getDatasetSequence().addDBRef(dna1xref);
+ DBRefEntry dna2xref = new DBRefEntry("UNIPROT", "ENSEMBL", "pep2",
+ new Mapping(mapfordna2));
+ dna2.getDatasetSequence().addDBRef(dna2xref);
+
+ /*
* execute method under test:
*/
AlignmentI cds = AlignmentUtils.makeCdsAlignment(new SequenceI[] {
* verify CDS has a dbref with mapping to peptide
*/
assertNotNull(cds1Dss.getDBRefs());
- assertEquals(1, cds1Dss.getDBRefs().length);
+ assertEquals(2, cds1Dss.getDBRefs().length);
dbref = cds1Dss.getDBRefs()[0];
- assertEquals("UNIPROT", dbref.getSource());
- assertEquals("0", dbref.getVersion());
- assertEquals("pep1", dbref.getAccessionId());
+ assertEquals(dna1xref.getSource(), dbref.getSource());
+ // version is via ensembl's primary ref
+ assertEquals(dna1xref.getVersion(), dbref.getVersion());
+ assertEquals(dna1xref.getAccessionId(), dbref.getAccessionId());
assertNotNull(dbref.getMap());
assertSame(pep1.getDatasetSequence(), dbref.getMap().getTo());
MapList cdsMapping = new MapList(new int[] { 1, 6 },
* verify peptide has added a dbref with reverse mapping to CDS
*/
assertNotNull(pep1.getDBRefs());
+ // FIXME pep1.getDBRefs() is 1 - is that the correct behaviour ?
assertEquals(2, pep1.getDBRefs().length);
dbref = pep1.getDBRefs()[1];
assertEquals("ENSEMBL", dbref.getSource());
assertEquals(7, m.getStart());
assertEquals(9, m.getEnd());
}
+
+ /**
+ * Test the method that just copies aligned sequences, provided all sequences
+ * to be aligned share the aligned sequence's dataset
+ */
+ @Test(groups = "Functional")
+ public void testAlignAsSameSequences()
+ {
+ SequenceI dna1 = new Sequence("dna1", "cccGGGTTTaaa");
+ SequenceI dna2 = new Sequence("dna2", "CCCgggtttAAA");
+ AlignmentI al1 = new Alignment(new SequenceI[] { dna1, dna2 });
+ ((Alignment) al1).createDatasetAlignment();
+
+ SequenceI dna3 = new Sequence(dna1);
+ SequenceI dna4 = new Sequence(dna2);
+ assertSame(dna3.getDatasetSequence(), dna1.getDatasetSequence());
+ assertSame(dna4.getDatasetSequence(), dna2.getDatasetSequence());
+ String seq1 = "-cc-GG-GT-TT--aaa";
+ dna3.setSequence(seq1);
+ String seq2 = "C--C-Cgg--gtt-tAA-A-";
+ dna4.setSequence(seq2);
+ AlignmentI al2 = new Alignment(new SequenceI[] { dna3, dna4 });
+ ((Alignment) al2).createDatasetAlignment();
+
+ assertTrue(AlignmentUtils.alignAsSameSequences(al1, al2));
+ assertEquals(seq1, al1.getSequenceAt(0).getSequenceAsString());
+ assertEquals(seq2, al1.getSequenceAt(1).getSequenceAsString());
+
+ /*
+ * add another sequence to 'aligned' - should still succeed, since
+ * unaligned sequences still share a dataset with aligned sequences
+ */
+ SequenceI dna5 = new Sequence("dna5", "CCCgggtttAAA");
+ dna5.createDatasetSequence();
+ al2.addSequence(dna5);
+ assertTrue(AlignmentUtils.alignAsSameSequences(al1, al2));
+ assertEquals(seq1, al1.getSequenceAt(0).getSequenceAsString());
+ assertEquals(seq2, al1.getSequenceAt(1).getSequenceAsString());
+
+ /*
+ * add another sequence to 'unaligned' - should fail, since now not
+ * all unaligned sequences share a dataset with aligned sequences
+ */
+ SequenceI dna6 = new Sequence("dna6", "CCCgggtttAAA");
+ dna6.createDatasetSequence();
+ al1.addSequence(dna6);
+ // JAL-2110 JBP Comment: what's the use case for this behaviour ?
+ assertFalse(AlignmentUtils.alignAsSameSequences(al1, al2));
+ }
+
+ @Test(groups = "Functional")
+ public void testAlignAsSameSequencesMultipleSubSeq()
+ {
+ SequenceI dna1 = new Sequence("dna1", "cccGGGTTTaaa");
+ SequenceI dna2 = new Sequence("dna2", "CCCgggtttAAA");
+ SequenceI as1 = dna1.deriveSequence();
+ SequenceI as2 = dna1.deriveSequence().getSubSequence(3, 7);
+ SequenceI as3 = dna2.deriveSequence();
+ as1.insertCharAt(6, 5, '-');
+ String s_as1 = as1.getSequenceAsString();
+ as2.insertCharAt(6, 5, '-');
+ String s_as2 = as2.getSequenceAsString();
+ as3.insertCharAt(6, 5, '-');
+ String s_as3 = as3.getSequenceAsString();
+ AlignmentI aligned = new Alignment(new SequenceI[] { as1, as2, as3 });
+
+ // why do we need to cast this still ?
+ ((Alignment) aligned).createDatasetAlignment();
+ SequenceI uas1 = dna1.deriveSequence();
+ SequenceI uas2 = dna1.deriveSequence().getSubSequence(3, 7);
+ SequenceI uas3 = dna2.deriveSequence();
+ AlignmentI tobealigned = new Alignment(new SequenceI[] { uas1, uas2,
+ uas3 });
+ ((Alignment) tobealigned).createDatasetAlignment();
+
+ assertTrue(AlignmentUtils.alignAsSameSequences(tobealigned, aligned));
+ assertEquals(s_as1, uas1.getSequenceAsString());
+ assertEquals(s_as2, uas2.getSequenceAsString());
+ assertEquals(s_as3, uas3.getSequenceAsString());
+ }
+
}
git://source.jalview.org / jalview.git / blobdiff