import java.util.Collections;
import java.util.Comparator;
import java.util.HashMap;
-import java.util.HashSet;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
-import java.util.Set;
/**
* Parses and writes features files, which may be in Jalview, GFF2 or GFF3
}
/**
- * Returns contents of a Jalview format features file
+ * Returns contents of a Jalview format features file, for visible features,
+ * as filtered by type and group. Features with a null group are displayed if
+ * their feature type is visible. Non-positional features may optionally be
+ * included (with no check on type or group).
*
* @param sequences
* source of features
* @param visible
* map of colour for each visible feature type
+ * @param visibleFeatureGroups
* @param includeNonPositional
* if true, include non-positional features (regardless of group or
* type)
* @return
*/
public String printJalviewFormat(SequenceI[] sequences,
- Map<String, FeatureColourI> visible, boolean includeNonPositional)
+ Map<String, FeatureColourI> visible,
+ List<String> visibleFeatureGroups, boolean includeNonPositional)
{
if (!includeNonPositional && (visible == null || visible.isEmpty()))
{
}
}
- // Work out which groups are both present and visible
- Set<String> groups = new HashSet<String>();
String[] types = visible == null ? new String[0] : visible.keySet()
.toArray(new String[visible.keySet().size()]);
- for (int i = 0; i < sequences.length; i++)
+ /*
+ * sort groups alphabetically, and ensure that null group is output last
+ */
+ List<String> sortedGroups = new ArrayList<String>(visibleFeatureGroups);
+ sortedGroups.remove(null);
+ Collections.sort(sortedGroups);
+ sortedGroups.add(null);
+
+ boolean foundSome = false;
+
+ /*
+ * first output any non-positional features
+ */
+ if (includeNonPositional)
{
- groups.addAll(sequences[i].getFeatures()
- .getFeatureGroups(true, types));
- if (includeNonPositional)
+ for (int i = 0; i < sequences.length; i++)
{
- groups.addAll(sequences[i].getFeatures().getFeatureGroups(false,
- types));
+ String sequenceName = sequences[i].getName();
+ for (SequenceFeature feature : sequences[i].getFeatures()
+ .getNonPositionalFeatures())
+ {
+ foundSome = true;
+ out.append(formatJalviewFeature(sequenceName, feature));
+ }
}
}
- /*
- * sort distinct groups so null group is output last
- */
- List<String> sortedGroups = new ArrayList<String>(groups);
- Collections.sort(sortedGroups, SORT_NULL_LAST);
-
- // TODO check where null group should be output
- boolean foundSome = false;
for (String group : sortedGroups)
{
if (group != null)
}
/*
- * output features within groups (non-positional first if wanted)
+ * output positional features within groups
*/
for (int i = 0; i < sequences.length; i++)
{
+ String sequenceName = sequences[i].getName();
List<SequenceFeature> features = new ArrayList<SequenceFeature>();
- if (includeNonPositional)
- {
- features.addAll(sequences[i].getFeatures().getFeaturesForGroup(
- false, group, types));
- }
if (types.length > 0)
{
features.addAll(sequences[i].getFeatures().getFeaturesForGroup(
for (SequenceFeature sequenceFeature : features)
{
- // we have features to output
foundSome = true;
- if (sequenceFeature.description == null
- || sequenceFeature.description.equals(""))
- {
- out.append(sequenceFeature.type).append(TAB);
- }
- else
- {
- if (sequenceFeature.links != null
- && sequenceFeature.getDescription().indexOf("<html>") == -1)
- {
- out.append("<html>");
- }
-
- out.append(sequenceFeature.description);
- if (sequenceFeature.links != null)
- {
- for (int l = 0; l < sequenceFeature.links.size(); l++)
- {
- String label = sequenceFeature.links.elementAt(l);
- String href = label.substring(label.indexOf("|") + 1);
- label = label.substring(0, label.indexOf("|"));
-
- if (sequenceFeature.description.indexOf(href) == -1)
- {
- out.append(" <a href=\"" + href + "\">" + label + "</a>");
- }
- }
-
- if (sequenceFeature.getDescription().indexOf("</html>") == -1)
- {
- out.append("</html>");
- }
- }
-
- out.append(TAB);
- }
- out.append(sequences[i].getName());
- out.append("\t-1\t");
- out.append(sequenceFeature.begin);
- out.append(TAB);
- out.append(sequenceFeature.end);
- out.append(TAB);
- out.append(sequenceFeature.type);
- if (!Float.isNaN(sequenceFeature.score))
- {
- out.append(TAB);
- out.append(sequenceFeature.score);
- }
- out.append(newline);
+ out.append(formatJalviewFeature(sequenceName, sequenceFeature));
}
}
}
/**
+ * @param out
+ * @param sequenceName
+ * @param sequenceFeature
+ */
+ protected String formatJalviewFeature(
+ String sequenceName, SequenceFeature sequenceFeature)
+ {
+ StringBuilder out = new StringBuilder(64);
+ if (sequenceFeature.description == null
+ || sequenceFeature.description.equals(""))
+ {
+ out.append(sequenceFeature.type).append(TAB);
+ }
+ else
+ {
+ if (sequenceFeature.links != null
+ && sequenceFeature.getDescription().indexOf("<html>") == -1)
+ {
+ out.append("<html>");
+ }
+
+ out.append(sequenceFeature.description);
+ if (sequenceFeature.links != null)
+ {
+ for (int l = 0; l < sequenceFeature.links.size(); l++)
+ {
+ String label = sequenceFeature.links.elementAt(l);
+ String href = label.substring(label.indexOf("|") + 1);
+ label = label.substring(0, label.indexOf("|"));
+
+ if (sequenceFeature.description.indexOf(href) == -1)
+ {
+ out.append(" <a href=\"" + href + "\">" + label + "</a>");
+ }
+ }
+
+ if (sequenceFeature.getDescription().indexOf("</html>") == -1)
+ {
+ out.append("</html>");
+ }
+ }
+
+ out.append(TAB);
+ }
+ out.append(sequenceName);
+ out.append("\t-1\t");
+ out.append(sequenceFeature.begin);
+ out.append(TAB);
+ out.append(sequenceFeature.end);
+ out.append(TAB);
+ out.append(sequenceFeature.type);
+ if (!Float.isNaN(sequenceFeature.score))
+ {
+ out.append(TAB);
+ out.append(sequenceFeature.score);
+ }
+ out.append(newline);
+
+ return out.toString();
+ }
+
+ /**
* Parse method that is called when a GFF file is dragged to the desktop
*/
@Override
* the sequences whose features are to be output
* @param visible
* a map whose keys are the type names of visible features
+ * @param visibleFeatureGroups
* @param includeNonPositionalFeatures
* @return
*/
public String printGffFormat(SequenceI[] sequences,
Map<String, FeatureColourI> visible,
+ List<String> visibleFeatureGroups,
boolean includeNonPositionalFeatures)
{
StringBuilder out = new StringBuilder(256);
- int version = gffVersion == 0 ? 2 : gffVersion;
- out.append(String.format("%s %d\n", GFF_VERSION, version));
- String source;
- boolean isnonpos;
+
+ out.append(String.format("%s %d\n", GFF_VERSION, gffVersion == 0 ? 2 : gffVersion));
+
+ if (!includeNonPositionalFeatures
+ && (visible == null || visible.isEmpty()))
+ {
+ return out.toString();
+ }
+
+ String[] types = visible == null ? new String[0] : visible.keySet()
+ .toArray(
+ new String[visible.keySet().size()]);
+
for (SequenceI seq : sequences)
{
- SequenceFeature[] features = seq.getSequenceFeatures();
- if (features != null)
+ List<SequenceFeature> features = new ArrayList<SequenceFeature>();
+ if (includeNonPositionalFeatures)
{
- for (SequenceFeature sf : features)
- {
- isnonpos = sf.begin == 0 && sf.end == 0;
- if (!includeNonPositionalFeatures && isnonpos)
- {
- /*
- * ignore non-positional features if not wanted
- */
- continue;
- }
- if (!isnonpos && !visible.containsKey(sf.type))
- {
- /*
- * ignore not visible features if not wanted
- */
- continue;
- }
+ features.addAll(seq.getFeatures().getNonPositionalFeatures());
+ }
+ if (visible != null && !visible.isEmpty())
+ {
+ features.addAll(seq.getFeatures().getPositionalFeatures(types));
+ }
- source = sf.featureGroup;
- if (source == null)
- {
- source = sf.getDescription();
- }
+ for (SequenceFeature sf : features)
+ {
+ String source = sf.featureGroup;
+ if (!sf.isNonPositional() && source != null
+ && !visibleFeatureGroups.contains(source))
+ {
+ // group is not visible
+ continue;
+ }
- out.append(seq.getName());
- out.append(TAB);
- out.append(source);
- out.append(TAB);
- out.append(sf.type);
- out.append(TAB);
- out.append(sf.begin);
- out.append(TAB);
- out.append(sf.end);
- out.append(TAB);
- out.append(sf.score);
- out.append(TAB);
-
- int strand = sf.getStrand();
- out.append(strand == 1 ? "+" : (strand == -1 ? "-" : "."));
- out.append(TAB);
-
- String phase = sf.getPhase();
- out.append(phase == null ? "." : phase);
-
- // miscellaneous key-values (GFF column 9)
- String attributes = sf.getAttributes();
- if (attributes != null)
- {
- out.append(TAB).append(attributes);
- }
+ if (source == null)
+ {
+ source = sf.getDescription();
+ }
- out.append(newline);
+ out.append(seq.getName());
+ out.append(TAB);
+ out.append(source);
+ out.append(TAB);
+ out.append(sf.type);
+ out.append(TAB);
+ out.append(sf.begin);
+ out.append(TAB);
+ out.append(sf.end);
+ out.append(TAB);
+ out.append(sf.score);
+ out.append(TAB);
+
+ int strand = sf.getStrand();
+ out.append(strand == 1 ? "+" : (strand == -1 ? "-" : "."));
+ out.append(TAB);
+
+ String phase = sf.getPhase();
+ out.append(phase == null ? "." : phase);
+
+ // miscellaneous key-values (GFF column 9)
+ String attributes = sf.getAttributes();
+ if (attributes != null)
+ {
+ out.append(TAB).append(attributes);
}
+
+ out.append(newline);
}
}
// rename sequences if GFF handler requested this
// TODO a more elegant way e.g. gffHelper.postProcess(newseqs) ?
- SequenceFeature[] sfs = seq.getSequenceFeatures();
- if (sfs != null)
+ List<SequenceFeature> sfs = seq.getFeatures().getPositionalFeatures();
+ if (!sfs.isEmpty())
{
- String newName = (String) sfs[0].getValue(GffHelperI.RENAME_TOKEN);
+ String newName = (String) sfs.get(0).getValue(
+ GffHelperI.RENAME_TOKEN);
if (newName != null)
{
seq.setName(newName);
import java.awt.Color;
import java.io.File;
import java.io.IOException;
+import java.util.ArrayList;
+import java.util.Arrays;
import java.util.HashMap;
+import java.util.List;
import java.util.Map;
import org.testng.annotations.BeforeClass;
*/
FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
Map<String, FeatureColourI> visible = fr.getDisplayedFeatureCols();
+ List<String> visibleGroups = new ArrayList<String>(
+ Arrays.asList(new String[] {}));
String exported = featuresFile.printJalviewFormat(
- al.getSequencesArray(), visible, false);
+ al.getSequencesArray(), visible, visibleGroups, false);
String expected = "No Features Visible";
assertEquals(expected, exported);
/*
* include non-positional features
*/
+ visibleGroups.add("uniprot");
exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
- visible, true);
- expected = "\nSTARTGROUP\tuniprot\nCath\tFER_CAPAA\t-1\t0\t0\tDomain\t0.0\nENDGROUP\tuniprot\n";
+ visible, visibleGroups, true);
+ expected = "Cath\tFER_CAPAA\t-1\t0\t0\tDomain\t0.0\n\nSTARTGROUP\tuniprot\nENDGROUP\tuniprot\n";
assertEquals(expected, exported);
/*
fr.setVisible("GAMMA-TURN");
visible = fr.getDisplayedFeatureCols();
exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
- visible, false);
+ visible, visibleGroups, false);
expected = "METAL\tcc9900\n"
+ "GAMMA-TURN\tff0000|00ffff|20.0|95.0|below|66.0\n"
+ "\nSTARTGROUP\tuniprot\n"
fr.setVisible("Pfam");
visible = fr.getDisplayedFeatureCols();
exported = featuresFile.printJalviewFormat(al.getSequencesArray(),
- visible, false);
+ visible, visibleGroups, false);
/*
* features are output within group, ordered by sequence and by type
*/
FeaturesFile featuresFile = new FeaturesFile();
FeatureRenderer fr = af.alignPanel.getFeatureRenderer();
Map<String, FeatureColourI> visible = new HashMap<String, FeatureColourI>();
- String exported = featuresFile.printGffFormat(
- al.getSequencesArray(), visible, false);
+ List<String> visibleGroups = new ArrayList<String>(
+ Arrays.asList(new String[] {}));
+ String exported = featuresFile.printGffFormat(al.getSequencesArray(),
+ visible, visibleGroups, false);
String gffHeader = "##gff-version 2\n";
assertEquals(gffHeader, exported);
exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
- true);
+ visibleGroups, true);
assertEquals(gffHeader, exported);
/*
.addSequenceFeature(
new SequenceFeature("GAMMA-TURN", "Turn", 36, 38, 2.1f,
"s3dm"));
- SequenceFeature sf = new SequenceFeature("Pfam", "", 20, 20, 0f, "Uniprot");
+ SequenceFeature sf = new SequenceFeature("Pfam", "", 20, 20, 0f,
+ "Uniprot");
sf.setAttributes("x=y;black=white");
sf.setStrand("+");
sf.setPhase("2");
* with no features displayed, exclude non-positional features
*/
exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
- false);
+ visibleGroups, false);
assertEquals(gffHeader, exported);
-
+
/*
* include non-positional features
*/
- exported = featuresFile.printGffFormat(al.getSequencesArray(),
- visible, true);
+ visibleGroups.add("Uniprot");
+ exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
+ visibleGroups, true);
String expected = gffHeader
+ "FER_CAPAA\tUniprot\tDomain\t0\t0\t0.0\t.\t.\n";
assertEquals(expected, exported);
-
+
/*
* set METAL (in uniprot group) and GAMMA-TURN visible, but not Pfam
+ * only Uniprot group visible here...
*/
fr.setVisible("METAL");
fr.setVisible("GAMMA-TURN");
visible = fr.getDisplayedFeatureCols();
- exported = featuresFile.printGffFormat(al.getSequencesArray(),
- visible, false);
+ exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
+ visibleGroups, 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);
+
+ /*
+ * set s3dm group visible
+ */
+ visibleGroups.add("s3dm");
+ exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
+ visibleGroups, 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";
assertEquals(expected, exported);
-
+
/*
* now set Pfam visible
*/
fr.setVisible("Pfam");
visible = fr.getDisplayedFeatureCols();
- exported = featuresFile.printGffFormat(al.getSequencesArray(),
- visible, false);
+ exported = featuresFile.printGffFormat(al.getSequencesArray(), visible,
+ visibleGroups, false);
// Pfam feature columns include strand(+), phase(2), attributes
- expected = gffHeader + "FER_CAPAA\tCath\tMETAL\t39\t39\t1.2\t.\t.\n"
+ 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";
assertEquals(expected, exported);
git://source.jalview.org / jalview.git / commitdiff