Searchable bigbed (#1447)

Add support for searching by bb extra indexes, including trix
igvteam · Nov 28, 2023 · cb9cd9b · cb9cd9b
1 parent c5114a9
commit cb9cd9b
Show file tree

Hide file tree

Showing 26 changed files with 337 additions and 225 deletions.
diff --git a/src/main/java/org/broad/igv/feature/genome/Genome.java b/src/main/java/org/broad/igv/feature/genome/Genome.java
@@ -257,6 +257,10 @@ private void addTracks(GenomeConfig config) {
                     res.setVisibilityWindow(vw);
                 }
 
+                if (trackConfig.searchTrix != null) {
+                    res.setTrixURL(trackConfig.searchTrix);
+                }
+
                 res.setFeatureInfoURL(trackConfig.infoURL);
                 Boolean indexed = trackConfig.indexed;
                 if (indexed != null) {
@@ -351,7 +355,7 @@ public Genome(String id, List<Chromosome> chromosomes) {
      * Return the canonical chromosome name for the (possibly) alias
      *
      * @param str chromosome or alias name
-     * @return the canonical chromsoome name -- i.e. chromosome name as defined by the reference sequence
+     * @return the canonical chromsoome name -if the chromosome exists.
      */
     public String getCanonicalChrName(String str) {
         if (str == null) {
@@ -361,9 +365,7 @@ public String getCanonicalChrName(String str) {
         } else if (chromAliasSource != null) {
             try {
                 ChromAlias aliasRecord = chromAliasSource.search(str);
-                if (aliasRecord == null) {
-                    return str;
-                } else {
+                if (aliasRecord != null) {
                     String chr = aliasRecord.getChr();
                     chrAliasCache.put(str, chr);
                     return chr;
@@ -487,9 +489,9 @@ public Chromosome getChromosome(String name) {
         if (chromosomeMap.containsKey(chrName)) {
             return chromosomeMap.get(chrName);
         } else {
-            int idx = this.chromosomeMap.size();
             int length = this.sequence.getChromosomeLength(chrName);
             if (length > 0) {
+                int idx = this.chromosomeMap.size();
                 Chromosome chromosome = new Chromosome(idx, chrName, length);
                 chromosomeMap.put(chrName, chromosome);
                 return chromosome;

diff --git a/src/main/java/org/broad/igv/feature/genome/GenomeManager.java b/src/main/java/org/broad/igv/feature/genome/GenomeManager.java
@@ -56,6 +56,7 @@
 import org.broad.igv.ui.PanelName;
 import org.broad.igv.ui.commandbar.GenomeListManager;
 import org.broad.igv.ui.panel.FrameManager;
+import org.broad.igv.ui.panel.ReferenceFrame;
 import org.broad.igv.ui.util.MessageUtils;
 import org.broad.igv.ui.util.ProgressBar;
 import org.broad.igv.ui.util.ProgressMonitor;
@@ -72,6 +73,7 @@
 import java.net.SocketException;
 import java.net.URL;
 import java.net.URLDecoder;
+import java.sql.Ref;
 import java.util.*;
 import java.util.List;
 
@@ -231,8 +233,9 @@ public Genome loadGenome(String genomePath, ProgressMonitor monitor) throws IOEx
 
             // hasInstance() test needed for unit tests
             if (IGV.hasInstance()) {
-                IGV.getInstance().goToLocus(newGenome.getDefaultPos());
+                IGV.getInstance().goToLocus(newGenome.getHomeChromosome()); //  newGenome.getDefaultPos());
                 loadGenomeAnnotations(newGenome);
+                IGV.getInstance().resetFrames();
             }
 
             if (PreferencesManager.getPreferences().getAsBoolean(Constants.CIRC_VIEW_ENABLED) && CircularViewUtilities.ping()) {

diff --git a/src/main/java/org/broad/igv/feature/genome/Sequence.java b/src/main/java/org/broad/igv/feature/genome/Sequence.java
@@ -38,13 +38,26 @@
  */
 public interface Sequence {
 
-    byte[] getSequence(String chr, int start, int end);
+    /**
+     * Return the sequence for the given range.  If sequence named "seq" does not exist returns null.
+     * @param chr
+     * @param start
+     * @param end
+     * @return  The sequence in bytes, or null if no sequence exists
+     */
+    byte[] getSequence(String seq, int start, int end) ;
 
-    byte getBase(String chr, int position);
+    byte getBase(String seq, int position);
 
     List<String> getChromosomeNames();
 
-    int getChromosomeLength(String chrname);
+    /**
+     * Return the given sequence length.  If no sequence exists with name "seq" return -1.
+     *
+     * @param seq
+     * @return
+     */
+    int getChromosomeLength(String seq);
 
     List<Chromosome> getChromosomes();
 

diff --git a/src/main/java/org/broad/igv/feature/genome/SequenceNotFoundException.java b/src/main/java/org/broad/igv/feature/genome/SequenceNotFoundException.java
@@ -0,0 +1,7 @@
+package org.broad.igv.feature.genome;
+
+public class SequenceNotFoundException extends RuntimeException {
+    public SequenceNotFoundException(String message) {
+        super(message);
+    }
+}
diff --git a/src/main/java/org/broad/igv/feature/genome/fasta/FastaIndexedSequence.java b/src/main/java/org/broad/igv/feature/genome/fasta/FastaIndexedSequence.java
@@ -191,7 +191,7 @@ public List<String> getChromosomeNames() {
 
     @Override
     public int getChromosomeLength(String chrname) {
-        return index.getSequenceSize(chrname);
+        return  index.getSequenceSize(chrname);
     }
 
     @Override

diff --git a/src/main/java/org/broad/igv/session/History.java b/src/main/java/org/broad/igv/session/History.java
@@ -31,6 +31,7 @@
 import org.broad.igv.ui.IGV;
 import org.broad.igv.ui.action.SearchCommand;
 import org.broad.igv.ui.panel.FrameManager;
+import org.broad.igv.util.LongRunningTask;
 
 import java.util.ArrayList;
 import java.util.LinkedList;
@@ -121,7 +122,7 @@ public void processItem(Entry entry) {
                 if (FrameManager.isGeneListMode()) {
                     IGV.getInstance().setGeneList(null, false);
                 }
-                (new SearchCommand(FrameManager.getDefaultFrame(), locus, false)).execute();
+                LongRunningTask.submit(new SearchCommand(FrameManager.getDefaultFrame(), locus, false));
                 //Zoom should be implicit in the locus
                 //FrameManager.getDefaultFrame().setZoom(entry.getZoom());
             }

diff --git a/src/main/java/org/broad/igv/track/FeatureSource.java b/src/main/java/org/broad/igv/track/FeatureSource.java
@@ -26,6 +26,7 @@
 package org.broad.igv.track;
 
 import htsjdk.tribble.Feature;
+import htsjdk.tribble.NamedFeature;
 import org.broad.igv.feature.LocusScore;
 import org.broad.igv.ui.panel.ReferenceFrame;
 
@@ -92,4 +93,17 @@ default void close() {
     default Object getHeader() {
         return null;
     }
+
+    /**
+     * Return true if the source can be searched for a feature by name
+     *
+     * @return
+     */
+    default boolean isSearchable() {
+        return false;
+    }
+
+    default NamedFeature search(String name) {
+        return null;
+    }
 }
diff --git a/src/main/java/org/broad/igv/track/FeatureTrack.java b/src/main/java/org/broad/igv/track/FeatureTrack.java
@@ -26,6 +26,7 @@
 package org.broad.igv.track;
 
 import htsjdk.tribble.Feature;
+import htsjdk.tribble.NamedFeature;
 import htsjdk.tribble.TribbleException;
 import org.broad.igv.event.IGVEvent;
 import org.broad.igv.logging.*;
@@ -1056,6 +1057,16 @@ public void marshalXML(Document document, Element element) {
 
     }
 
+    @Override
+    public boolean isSearchable() {
+        return source.isSearchable();
+    }
+
+    @Override
+    public NamedFeature search(String token) {
+        return source.search(token);
+    }
+
     @Override
     public void unmarshalXML(Element element, Integer version) {
 

diff --git a/src/main/java/org/broad/igv/track/Track.java b/src/main/java/org/broad/igv/track/Track.java
@@ -31,6 +31,7 @@
 
 
 import htsjdk.tribble.Feature;
+import htsjdk.tribble.NamedFeature;
 import org.broad.igv.renderer.ContinuousColorScale;
 import org.broad.igv.renderer.DataRange;
 import org.broad.igv.renderer.Renderer;
@@ -272,9 +273,12 @@ default Color getExplicitAltColor() {
 
     void setAutoScale(boolean autoScale);
 
-    default boolean isShowFeatureNames() {return true;}
+    default boolean isShowFeatureNames() {
+        return true;
+    }
 
-    default void setShowFeatureNames(boolean b) {}
+    default void setShowFeatureNames(boolean b) {
+    }
 
     /**
      * Return the java property or attribute for the feature display name.  Default is "null", in which case the
@@ -286,6 +290,19 @@ default String getLabelField() {
         return null;
     }
 
+    /**
+     * Return true if the track can be searched for a feature by name.
+     *
+     * @return
+     */
+    default boolean isSearchable() {
+        return false;
+    }
+
+    default NamedFeature search(String token) {
+        return null;
+    }
+
     default void repaint() {
         IGV.getInstance().repaint(this);
     }

diff --git a/src/main/java/org/broad/igv/track/TrackLoader.java b/src/main/java/org/broad/igv/track/TrackLoader.java
@@ -76,6 +76,7 @@
 import org.broad.igv.sam.reader.IndexNotFoundException;
 import org.broad.igv.tdf.TDFDataSource;
 import org.broad.igv.tdf.TDFReader;
+import org.broad.igv.ucsc.Trix;
 import org.broad.igv.ucsc.bb.BBDataSource;
 import org.broad.igv.ucsc.bb.BBFeatureSource;
 import org.broad.igv.ucsc.bb.BBFile;
@@ -792,9 +793,9 @@ public void loadBWFile(ResourceLocator locator, List<Track> newTracks, Genome ge
         String trackId = locator.getPath();
 
         String path = locator.getPath();
-        BBFile reader = new BBFile(path, genome);
-        Track track = null;
-
+        String trixURL = locator.getTrixURL();
+        BBFile reader = trixURL == null ? new BBFile(path, genome) : new BBFile(path, genome, trixURL);
+        Track track;
         if (reader.isBigWigFile()) {
             BBDataSource bigwigSource = new BBDataSource(reader, genome);
             track = new DataSourceTrack(locator, trackId, trackName, bigwigSource);

diff --git a/src/main/java/org/broad/igv/ucsc/Trix.java b/src/main/java/org/broad/igv/ucsc/Trix.java
@@ -111,7 +111,8 @@ String _getBuffer(String searchWord) throws IOException {
         int end = 65536;
         List<IndexEntry> indexes = this.getIndex();
         for (IndexEntry entry : indexes) {
-            String trimmedKey = entry.key.substring(0, searchWord.length());
+            int trimEnd = Math.min(entry.key.length(), searchWord.length());
+            String trimmedKey = entry.key.substring(0, trimEnd);
             if (trimmedKey.compareTo(searchWord) < 0) {
                 start = entry.value;
                 end = entry.value + 65536;

diff --git a/src/main/java/org/broad/igv/ucsc/bb/BBFeatureSource.java b/src/main/java/org/broad/igv/ucsc/bb/BBFeatureSource.java
@@ -25,9 +25,12 @@
 
 package org.broad.igv.ucsc.bb;
 
+import htsjdk.tribble.NamedFeature;
 import org.broad.igv.feature.BasicFeature;
 import org.broad.igv.feature.LocusScore;
 import org.broad.igv.feature.genome.Genome;
+import org.broad.igv.logging.LogManager;
+import org.broad.igv.logging.Logger;
 import org.broad.igv.track.FeatureSource;
 import org.broad.igv.track.WindowFunction;
 
@@ -42,10 +45,9 @@
  */
 public class BBFeatureSource implements FeatureSource {
 
-    final int screenWidth = 1000; // TODO use actual screen width
+    private static Logger log = LogManager.getLogger(BBFeatureSource.class);
     private final Genome genome;
 
-
     Collection<WindowFunction> availableWindowFunctions =
             Arrays.asList(WindowFunction.min, WindowFunction.mean, WindowFunction.max, WindowFunction.none);
 
@@ -56,9 +58,6 @@ public class BBFeatureSource implements FeatureSource {
 
     private Map<WindowFunction, List<LocusScore>> wholeGenomeScores;
 
-    // Lookup table to support chromosome aliasing.
-    private Map<String, String> chrNameMap = new HashMap();
-
     public BBFeatureSource(BBFile reader, Genome genome) throws IOException {
 
         super();
@@ -108,6 +107,20 @@ public List<LocusScore> getCoverageScores(String chr, int start, int end, int zo
         return null;
     }
 
+    public boolean isSearchable() {
+        return reader.isSearchable();
+    }
+
+    @Override
+    public NamedFeature search(String term) {
+        try {
+            return reader.search(term);
+        } catch (IOException e) {
+            log.error("Error searching for: " + term, e);
+            return null;
+        }
+    }
+
     static class FeatureIterator implements Iterator<BasicFeature> {
 
         List<BasicFeature> features;
@@ -125,10 +138,9 @@ public FeatureIterator(List<BasicFeature> features, int start, int end) {
         }
 
         private void advance() {
-            if(idx == features.size()) {
+            if (idx == features.size()) {
                 next = null;
-            }
-            else {
+            } else {
                 while (idx < features.size()) {
                     next = features.get(idx++);
                     if (next.getStart() > end) {

diff --git a/src/main/java/org/broad/igv/ucsc/bb/BBFile.java b/src/main/java/org/broad/igv/ucsc/bb/BBFile.java
@@ -1,6 +1,7 @@
 package org.broad.igv.ucsc.bb;
 
 import htsjdk.samtools.seekablestream.SeekableStream;
+import htsjdk.tribble.NamedFeature;
 import org.broad.igv.data.BasicScore;
 import org.broad.igv.feature.BasicFeature;
 import org.broad.igv.feature.LocusScore;
@@ -112,6 +113,9 @@ public Set<String> getChromosomeNames() {
         return chrNames;
     }
 
+    public void setTrix(Trix trix) {
+    }
+
 
     enum Type {BIGWIG, BIGBED}
 
@@ -384,6 +388,9 @@ BBZoomHeader zoomLevelForScale(double bpPerPixel) {
         return lastLevel.reductionLevel / 2 < bpPerPixel ? lastLevel : null;
     }
 
+    public boolean isSearchable() {
+        return header.extraIndexCount > 0;
+    }
 
     /**
      * Search the extended BP tree for the search term, and return any matching features.  This only works
@@ -394,6 +401,7 @@ BBZoomHeader zoomLevelForScale(double bpPerPixel) {
      * @param term
      * @returns {Promise<void>}
      */
+
     public BasicFeature search(String term) throws IOException {
 
         if (this.header == null) {
@@ -403,6 +411,16 @@ public BasicFeature search(String term) throws IOException {
             return null;
         }
 
+        if (this.trix != null) {
+            String termLower = term.toLowerCase();
+            Map<String, String[]> results = trix.search(termLower);
+            if (results != null) {
+                String[] exactMatches = results.get(termLower);
+                if (exactMatches.length > 0) term = exactMatches[0];
+            }
+        }
+
+
         long[] region = this.searchForRegions(term);  // Either 1 or no (undefined) reginos returned for now
         if (region != null) {
             long start = region[0];
@@ -412,11 +430,18 @@ public BasicFeature search(String term) throws IOException {
                 is.seek(start);
                 is.readFully(buffer);
                 List<BasicFeature> features = decodeFeatures(buffer, -1, -1, -1);
-                BasicFeature largest = features.stream().reduce((f1, f2) -> {
+
+                // Filter features to those matching term
+                final String searchTerm = term;
+
+                BasicFeature largest = features.stream().filter(f -> {
+                    return f.getName().equalsIgnoreCase(searchTerm) || f.getAttributes().values().stream().anyMatch(v -> v.equalsIgnoreCase(searchTerm));
+                }).reduce((f1, f2) -> {
                     int l1 = f1.getEnd() - f1.getStart();
                     int l2 = f2.getEnd() - f2.getStart();
                     return l1 > l2 ? f1 : f2;
                 }).get();
+
                 return largest;
             }
         }