import static org.testng.AssertJUnit.assertTrue;
import static org.testng.internal.junit.ArrayAsserts.assertArrayEquals;
+import java.awt.Color;
+import java.io.File;
+import java.io.IOException;
+import java.util.HashMap;
+import java.util.Iterator;
+import java.util.List;
+import java.util.Map;
+
+import org.testng.annotations.AfterClass;
+import org.testng.annotations.BeforeClass;
+import org.testng.annotations.Test;
+
import jalview.api.FeatureColourI;
import jalview.api.FeatureRenderer;
import jalview.datamodel.Alignment;
import jalview.util.matcher.Condition;
import jalview.viewmodel.seqfeatures.FeatureRendererModel;
import jalview.viewmodel.seqfeatures.FeatureRendererModel.FeatureSettingsBean;
-
-import java.awt.Color;
-import java.io.File;
-import java.io.IOException;
-import java.util.HashMap;
-import java.util.Iterator;
-import java.util.List;
-import java.util.Map;
-
-import org.testng.annotations.AfterClass;
-import org.testng.annotations.BeforeClass;
-import org.testng.annotations.Test;
+import junit.extensions.PA;
public class FeaturesFileTest
{
.getFeatureColours();
FeaturesFile featuresFile = new FeaturesFile(
"examples/exampleFeatures.txt", DataSourceType.FILE);
- assertTrue("Test " + "Features file test"
- + "\nFailed to parse features file.",
+ assertTrue(
+ "Test " + "Features file test"
+ + "\nFailed to parse features file.",
featuresFile.parse(al.getDataset(), colours, true));
/*
Map<String, FeatureColourI> colours = af.getFeatureRenderer()
.getFeatureColours();
// GFF2 uses space as name/value separator in column 9
- String gffData = "METAL\tcc9900\n"
- + "GFF\n"
+ String gffData = "METAL\tcc9900\n" + "GFF\n"
+ "FER_CAPAA\tuniprot\tMETAL\t44\t45\t4.0\t.\t.\tNote Iron-sulfur; Note 2Fe-2S\n"
+ "FER1_SOLLC\tuniprot\tPfam\t55\t130\t2.0\t.\t.";
FeaturesFile featuresFile = new FeaturesFile(gffData,
AlignFrame af = new AlignFrame(al, 500, 500);
Map<String, FeatureColourI> colours = af.getFeatureRenderer()
.getFeatureColours();
- // GFF3 uses '=' separator for name/value pairs in colum 9
+ // GFF3 uses '=' separator for name/value pairs in column 9
+ // comma (%2C) equals (%3D) or semi-colon (%3B) should be url-escaped in
+ // values
String gffData = "##gff-version 3\n"
+ "FER_CAPAA\tuniprot\tMETAL\t39\t39\t0.0\t.\t.\t"
- + "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465\n"
+ + "Note=Iron-sulfur (2Fe-2S);Note=another note,and another;evidence=ECO%3B0000255%2CPROSITE%3DProRule:PRU00465;"
+ + "CSQ=AF=21,POLYPHEN=benign,possibly_damaging,clin_sig=Benign%3Dgood\n"
+ "FER1_SOLLC\tuniprot\tPfam\t55\t130\t3.0\t.\t.\tID=$23";
FeaturesFile featuresFile = new FeaturesFile(gffData,
DataSourceType.PASTE);
assertEquals(1, sfs.size());
SequenceFeature sf = sfs.get(0);
// description parsed from Note attribute
- assertEquals("Iron-sulfur (2Fe-2S),another note", sf.description);
+ assertEquals("Iron-sulfur (2Fe-2S),another note,and another",
+ sf.description);
assertEquals(39, sf.begin);
assertEquals(39, sf.end);
assertEquals("uniprot", sf.featureGroup);
assertEquals("METAL", sf.type);
- assertEquals(
- "Note=Iron-sulfur (2Fe-2S);Note=another note;evidence=ECO:0000255|PROSITE-ProRule:PRU00465",
- sf.getValue("ATTRIBUTES"));
+ assertEquals(5, sf.otherDetails.size());
+ assertEquals("ECO;0000255,PROSITE=ProRule:PRU00465", // url decoded
+ sf.getValue("evidence"));
+ assertEquals("Iron-sulfur (2Fe-2S),another note,and another",
+ sf.getValue("Note"));
+ assertEquals("21", sf.getValueAsString("CSQ", "AF"));
+ assertEquals("benign,possibly_damaging",
+ sf.getValueAsString("CSQ", "POLYPHEN"));
+ assertEquals("Benign=good", sf.getValueAsString("CSQ", "clin_sig")); // url
+ // decoded
+ // todo change STRAND and !Phase into fields of SequenceFeature instead
+ assertEquals(".", sf.otherDetails.get("STRAND"));
+ assertEquals(0, sf.getStrand());
+ assertEquals(".", sf.getPhase());
// verify feature on FER1_SOLLC1
sfs = al.getSequenceAt(2).getDatasetSequence().getSequenceFeatures();
SequenceI seq2 = dataset.findName("seq2");
assertNotNull(seq1);
assertNotNull(seq2);
- assertFalse(
- "Failed to replace dummy seq1 with real sequence",
+ assertFalse("Failed to replace dummy seq1 with real sequence",
seq1 instanceof SequenceDummy
&& ((SequenceDummy) seq1).isDummy());
- assertFalse(
- "Failed to replace dummy seq2 with real sequence",
+ assertFalse("Failed to replace dummy seq2 with real sequence",
seq2 instanceof SequenceDummy
&& ((SequenceDummy) seq2).isDummy());
String placeholderseq = new SequenceDummy("foo").getSequenceAsString();
assertFalse("dummy replacement buggy for seq2",
placeholderseq.equals(seq2.getSequenceAsString()));
assertNotNull("No features added to seq1", seq1.getSequenceFeatures());
- assertEquals("Wrong number of features", 3, seq1.getSequenceFeatures()
- .size());
+ assertEquals("Wrong number of features", 3,
+ seq1.getSequenceFeatures().size());
assertTrue(seq2.getSequenceFeatures().isEmpty());
- assertEquals(
- "Wrong number of features",
- 0,
- seq2.getSequenceFeatures() == null ? 0 : seq2
- .getSequenceFeatures().size());
- assertTrue(
- "Expected at least one CDNA/Protein mapping for seq1",
+ assertEquals("Wrong number of features", 0,
+ seq2.getSequenceFeatures() == null ? 0
+ : seq2.getSequenceFeatures().size());
+ assertTrue("Expected at least one CDNA/Protein mapping for seq1",
dataset.getCodonFrame(seq1) != null
&& dataset.getCodonFrame(seq1).size() > 0);
AlignmentI dataset = new Alignment(new SequenceI[] {});
FeaturesFile ffile = new FeaturesFile(simpleGffFile,
DataSourceType.FILE);
-
+
boolean parseResult = ffile.parse(dataset, null, false, false);
assertTrue("return result should be true", parseResult);
checkDatasetfromSimpleGff3(dataset);
@Test(groups = { "Functional" })
public void simpleGff3FileLoader() throws IOException
{
- AlignFrame af = new FileLoader(false).LoadFileWaitTillLoaded(
- simpleGffFile, DataSourceType.FILE);
+ AlignFrame af = new FileLoader(false)
+ .LoadFileWaitTillLoaded(simpleGffFile, DataSourceType.FILE);
assertTrue(
"Didn't read the alignment into an alignframe from Gff3 File",
af != null);
AlignmentI dataset = new Alignment(new SequenceI[] {});
FeaturesFile ffile = new FeaturesFile(simpleGffFile,
DataSourceType.FILE);
-
+
boolean parseResult = ffile.parse(dataset, null, false, true);
assertTrue("return result (relaxedID matching) should be true",
parseResult);
.getFeatureColours();
String features = "METAL\tcc9900\n"
+ "GAMMA-TURN\tred|0,255,255|20.0|95.0|below|66.0\n"
- + "Pfam\tred\n"
- + "STARTGROUP\tuniprot\n"
+ + "Pfam\tred\n" + "STARTGROUP\tuniprot\n"
+ "Cath\tFER_CAPAA\t-1\t0\t0\tDomain\n" // non-positional feature
+ "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\n"
+ "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\n"
* empty feature group to check handled correctly
*/
SequenceI seq = al.getSequenceAt(1); // FER_CAPAN
- seq.addSequenceFeature(new SequenceFeature("Pfam", "desc1", 0, 0, 1.3f,
- null));
- seq.addSequenceFeature(new SequenceFeature("Pfam", "desc2", 4, 9,
- Float.NaN, null));
+ seq.addSequenceFeature(
+ new SequenceFeature("Pfam", "desc1", 0, 0, 1.3f, null));
+ seq.addSequenceFeature(
+ new SequenceFeature("Pfam", "desc2", 4, 9, Float.NaN, null));
seq = al.getSequenceAt(2); // FER1_SOLLC
- seq.addSequenceFeature(new SequenceFeature("Pfam", "desc3", 0, 0,
- Float.NaN, ""));
- seq.addSequenceFeature(new SequenceFeature("Pfam", "desc4", 5, 8,
- -2.6f, ""));
+ seq.addSequenceFeature(
+ new SequenceFeature("Pfam", "desc3", 0, 0, Float.NaN, ""));
+ seq.addSequenceFeature(
+ new SequenceFeature("Pfam", "desc4", 5, 8, -2.6f, ""));
/*
* first with no features displayed, exclude non-positional features
*/
FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
String exported = featuresFile
- .printJalviewFormat(al.getSequencesArray(), fr, false);
+ .printJalviewFormat(al.getSequencesArray(), fr, false, false);
String expected = "No Features Visible";
assertEquals(expected, exported);
*/
fr.setGroupVisibility("uniprot", true);
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- true);
+ true, false);
expected = "\nSTARTGROUP\tuniprot\n"
+ "Cath\tFER_CAPAA\t-1\t0\t0\tDomain\t0.0\n"
+ "ENDGROUP\tuniprot\n\n"
fr.setVisible("METAL");
fr.setVisible("GAMMA-TURN");
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
expected = "METAL\tcc9900\n"
+ "GAMMA-TURN\tscore|ff0000|00ffff|noValueMin|20.0|95.0|below|66.0\n"
+ "\nSTARTGROUP\tuniprot\n"
- + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
+ "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
+ + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
+ "ENDGROUP\tuniprot\n";
assertEquals(expected, exported);
*/
fr.setVisible("Pfam");
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
/*
* features are output within group, ordered by sequence and type
*/
- expected = "METAL\tcc9900\n"
- + "Pfam\tff0000\n"
+ expected = "METAL\tcc9900\n" + "Pfam\tff0000\n"
+ "GAMMA-TURN\tscore|ff0000|00ffff|noValueMin|20.0|95.0|below|66.0\n"
+ "\nSTARTGROUP\tuniprot\n"
- + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
+ "Iron\tFER_CAPAA\t-1\t39\t39\tMETAL\t0.0\n"
+ "<html>Pfam domain<a href=\"http://pfam.xfam.org/family/PF00111\">Pfam_3_4</a></html>\tFER_CAPAA\t-1\t20\t20\tPfam\t0.0\n"
+ + "Turn\tFER_CAPAA\t-1\t36\t38\tGAMMA-TURN\t0.0\n"
+ "ENDGROUP\tuniprot\n"
// null / empty group features are output after named groups
+ "\ndesc2\tFER_CAPAN\t-1\t4\t9\tPfam\n"
+ "\ndesc2\tFER_CAPAN\t-1\t4\t9\tPfam\n"
+ "\ndesc4\tFER1_SOLLC\t-1\t5\t8\tPfam\t-2.6\n";
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
assertEquals(expected, exported);
/*
* include non-positional (overrides group not shown)
*/
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- true);
+ true, false);
expected = "METAL\tcc9900\n" + "Pfam\tff0000\n"
+ "GAMMA-TURN\tscore|ff0000|00ffff|noValueMin|20.0|95.0|below|66.0\n"
+ "\nSTARTGROUP\tuniprot\n"
FeatureRendererModel fr = (FeatureRendererModel) af.alignPanel
.getFeatureRenderer();
String exported = featuresFile.printGffFormat(al.getSequencesArray(),
- fr, false);
+ fr, false, false);
String gffHeader = "##gff-version 2\n";
assertEquals(gffHeader, exported);
- exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- true);
+ exported = featuresFile.printGffFormat(al.getSequencesArray(), fr, true,
+ false);
assertEquals(gffHeader, exported);
/*
new SequenceFeature("Domain", "Cath", 0, 0, 0f, "Uniprot"));
al.getSequenceAt(0).addSequenceFeature(
new SequenceFeature("METAL", "Cath", 39, 39, 1.2f, null));
- al.getSequenceAt(1)
- .addSequenceFeature(
- new SequenceFeature("GAMMA-TURN", "Turn", 36, 38, 2.1f,
- "s3dm"));
+ al.getSequenceAt(1).addSequenceFeature(new SequenceFeature("GAMMA-TURN",
+ "Turn", 36, 38, 2.1f, "s3dm"));
SequenceFeature sf = new SequenceFeature("Pfam", "", 20, 20, 0f,
"Uniprot");
- sf.setAttributes("x=y;black=white");
sf.setStrand("+");
sf.setPhase("2");
+ sf.setValue("x", "y");
+ sf.setValue("black", "white");
+ Map<String, String> csq = new HashMap<>();
+ csq.put("SIFT", "benign,mostly benign,cloudy, with meatballs");
+ csq.put("consequence", "missense_variant");
+ sf.setValue("CSQ", csq);
al.getSequenceAt(1).addSequenceFeature(sf);
/*
* with no features displayed, exclude non-positional features
*/
exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
assertEquals(gffHeader, exported);
/*
*/
fr.setGroupVisibility("Uniprot", true);
fr.setGroupVisibility("s3dm", false);
- exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- true);
+ exported = featuresFile.printGffFormat(al.getSequencesArray(), fr, true,
+ false);
String expected = gffHeader
+ "FER_CAPAA\tUniprot\tDomain\t0\t0\t0.0\t.\t.\n";
assertEquals(expected, exported);
fr.setVisible("METAL");
fr.setVisible("GAMMA-TURN");
exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
// METAL feature has null group: description used for column 2
expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n";
assertEquals(expected, exported);
*/
fr.setGroupVisibility("s3dm", true);
exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
// METAL feature has null group: description used for column 2
expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
+ "FER_CAPAN\ts3dm\tGAMMA-TURN\t36\t38\t2.1\t.\t.\n";
*/
fr.setVisible("Pfam");
exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
// Pfam feature columns include strand(+), phase(2), attributes
- expected = gffHeader
- + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
- + "FER_CAPAN\ts3dm\tGAMMA-TURN\t36\t38\t2.1\t.\t.\n"
- + "FER_CAPAN\tUniprot\tPfam\t20\t20\t0.0\t+\t2\tx=y;black=white\n";
+ expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
+ // CSQ output as CSQ=att1=value1,att2=value2
+ // note all commas are encoded here which is wrong - it should be
+ // SIFT=benign,mostly benign,cloudy%2C with meatballs
+ + "FER_CAPAN\tUniprot\tPfam\t20\t20\t0.0\t+\t2\tx=y;black=white;"
+ + "CSQ=SIFT=benign%2Cmostly benign%2Ccloudy%2C with meatballs,consequence=missense_variant\n"
+ + "FER_CAPAN\ts3dm\tGAMMA-TURN\t36\t38\t2.1\t.\t.\n";
assertEquals(expected, exported);
}
assertTrue(matcher.isByScore());
assertSame(matcher.getMatcher().getCondition(), Condition.LT);
assertEquals(matcher.getMatcher().getPattern(), "1.3");
- assertEquals(matcher.getMatcher().getFloatValue(), 1.3f);
+ assertEquals(PA.getValue(matcher.getMatcher(), "floatValue"), 1.3f);
assertFalse(matchers.hasNext());
}
sf2.setValue("clin_sig", "Benign");
sf2.setValue("AF", "46");
al.getSequenceAt(0).addSequenceFeature(sf2);
-
+
FeaturesFile featuresFile = new FeaturesFile();
FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
final String gffHeader = "##gff-version 2\n";
fr.setVisible("METAL");
fr.setColour("METAL", new FeatureColour(Color.PINK));
String exported = featuresFile.printGffFormat(al.getSequencesArray(),
- fr, false);
+ fr, false, false);
String expected = gffHeader
- + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
- + "FER_CAPAA\tCath\tMETAL\t41\t41\t0.6\t.\t.\n";
+ + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\tclin_sig=Likely Pathogenic;AF=24\n"
+ + "FER_CAPAA\tCath\tMETAL\t41\t41\t0.6\t.\t.\tclin_sig=Benign;AF=46\n";
assertEquals(expected, exported);
/*
* now threshold to Score > 1.1 - should exclude sf2
*/
- FeatureColourI fc = new FeatureColour(Color.white, Color.BLACK,
+ FeatureColourI fc = new FeatureColour(null, Color.white, Color.BLACK,
Color.white, 0f, 2f);
fc.setAboveThreshold(true);
fc.setThreshold(1.1f);
fr.setColour("METAL", fc);
exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- false);
- expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n";
+ false, false);
+ expected = gffHeader
+ + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\tclin_sig=Likely Pathogenic;AF=24\n";
assertEquals(expected, exported);
/*
*/
fc.setAboveThreshold(false);
exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- false);
- expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
- + "FER_CAPAA\tCath\tMETAL\t41\t41\t0.6\t.\t.\n";
+ false, false);
+ expected = gffHeader
+ + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\tclin_sig=Likely Pathogenic;AF=24\n"
+ + "FER_CAPAA\tCath\tMETAL\t41\t41\t0.6\t.\t.\tclin_sig=Benign;AF=46\n";
assertEquals(expected, exported);
/*
"clin_sig"));
fr.setFeatureFilter("METAL", filter);
exported = featuresFile.printGffFormat(al.getSequencesArray(), fr,
- false);
- expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t41\t41\t0.6\t.\t.\n";
+ false, false);
+ expected = gffHeader
+ + "FER_CAPAA\tCath\tMETAL\t41\t41\t0.6\t.\t.\tclin_sig=Benign;AF=46\n";
assertEquals(expected, exported);
}
sf2.setValue("clin_sig", "Benign");
sf2.setValue("AF", "46");
al.getSequenceAt(0).addSequenceFeature(sf2);
-
+
FeaturesFile featuresFile = new FeaturesFile();
FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
fr.findAllFeatures(true);
-
+
fr.setVisible("METAL");
fr.setColour("METAL", new FeatureColour(Color.PINK));
- String exported = featuresFile.printJalviewFormat(
- al.getSequencesArray(),
- fr, false);
+ String exported = featuresFile
+ .printJalviewFormat(al.getSequencesArray(), fr, false, false);
String expected = "METAL\tffafaf\n\nSTARTGROUP\tgrp1\n"
+ "Cath\tFER_CAPAA\t-1\t39\t39\tMETAL\t1.2\n"
+ "ENDGROUP\tgrp1\n\nSTARTGROUP\tgrp2\n"
+ "Cath\tFER_CAPAA\t-1\t41\t41\tMETAL\t0.6\n"
+ "ENDGROUP\tgrp2\n";
assertEquals(expected, exported);
-
+
/*
* now threshold to Score > 1.1 - should exclude sf2
* (and there should be no empty STARTGROUP/ENDGROUP output)
*/
- FeatureColourI fc = new FeatureColour(Color.white, Color.BLACK,
+ FeatureColourI fc = new FeatureColour(null, Color.white, Color.BLACK,
Color.white, 0f, 2f);
fc.setAboveThreshold(true);
fc.setThreshold(1.1f);
fr.setColour("METAL", fc);
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
expected = "METAL\tscore|ffffff|000000|noValueMin|abso|0.0|2.0|above|1.1\n\n"
+ "STARTGROUP\tgrp1\n"
+ "Cath\tFER_CAPAA\t-1\t39\t39\tMETAL\t1.2\n"
+ "ENDGROUP\tgrp1\n";
assertEquals(expected, exported);
-
+
/*
* remove threshold and check sf2 is exported
*/
fc.setAboveThreshold(false);
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
expected = "METAL\tscore|ffffff|000000|noValueMin|abso|0.0|2.0|none\n\n"
+ "STARTGROUP\tgrp1\n"
+ "Cath\tFER_CAPAA\t-1\t39\t39\tMETAL\t1.2\n"
+ "Cath\tFER_CAPAA\t-1\t41\t41\tMETAL\t0.6\n"
+ "ENDGROUP\tgrp2\n";
assertEquals(expected, exported);
-
+
/*
* filter on (clin_sig contains Benign) - should include sf2 and exclude sf1
*/
"clin_sig"));
fr.setFeatureFilter("METAL", filter);
exported = featuresFile.printJalviewFormat(al.getSequencesArray(), fr,
- false);
+ false, false);
expected = "FER_CAPAA\tCath\tMETAL\t41\t41\t0.6\t.\t.\n";
expected = "METAL\tscore|ffffff|000000|noValueMin|abso|0.0|2.0|none\n\n"
+ "STARTFILTERS\nMETAL\tclin_sig Contains benign\nENDFILTERS\n\n"
git://source.jalview.org / jalview.git / blobdiff