O(N^2) -> O(N) loop for a big speedup

hammerlab · Mar 18, 2015 · 616c3c1 · 616c3c1
1 parent 9f7d45e
commit 616c3c1
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 12 deletions.
diff --git a/lib/underscore.js b/lib/underscore.js
@@ -14,5 +14,8 @@ declare module "underscore" {
 
   declare function chain<S>(obj: S): any;
   declare function any<T>(list: Array<T>, pred: (el: T)=>boolean): boolean;
+
+  declare function each<T>(o: {[key:string]: T}, iteratee: (val: T, key: string)=>void): void;
+  declare function each<T>(a: T[], iteratee: (val: T, key: string)=>void): void;
 }
 
diff --git a/src/BigBedDataSource.js b/src/BigBedDataSource.js
@@ -76,20 +76,26 @@ function parseBedFeature(f): Gene {
 
 function createBigBedDataSource(remoteSource: BigBed): BigBedSource {
   // Collection of genes that have already been loaded.
-  var genes: Array<Gene> = [];
+  var genes: {[key:string]: Gene} = {};
 
   // Ranges for which we have complete information -- no need to hit network.
   var coveredRanges: Array<ContigInterval<string>> = []
 
   function addGene(newGene) {
-    if (!_.findWhere(genes, {id: newGene.id})) {
-      genes.push(newGene);
+    if (!genes[newGene.id]) {
+      genes[newGene.id] = newGene;
     }
   }
 
   function getGenesInRange(range: ContigInterval<string>): Gene[] {
     if (!range) return [];
-    return genes.filter(gene => range.intersects(gene.position));
+    var results = [];
+    _.each(genes, gene => {
+      if (range.intersects(gene.position)) {
+        results.push(gene);
+      }
+    });
+    return results;
   }
 
   function fetch(range: GenomeRange) {

diff --git a/test/BigBedDataSource-test.js b/test/BigBedDataSource-test.js
@@ -17,14 +17,8 @@ describe('BigBedDataSource', function() {
     return createBigBedDataSource(new BigBed('/test/data/ensembl.chr17.bb'));
   }
 
-  // Shorthand
-  function ci(contig, start, stop) {
-    return new ContigInterval(contig, start, stop);
-  }
-
   it('should extract features in a range', function(done) {
     this.timeout(5000);
-    console.time('extract');
     var source = getTestSource();
 
     // No genes fetched initially
@@ -34,7 +28,6 @@ describe('BigBedDataSource', function() {
 
     // Fetching that one gene should cache its entire block.
     source.on('newdata', () => {
-      console.timeEnd('extract');
       var tp53s = source.getGenesInRange(tp53range);
       expect(tp53s).to.have.length(1);
 
@@ -43,6 +36,10 @@ describe('BigBedDataSource', function() {
       expect(tp53.exons).to.have.length(11);
       done();
     });
-    source.rangeChanged(tp53range);
+    source.rangeChanged({
+      contig: tp53range.contig,
+      start: tp53range.start(),
+      stop: tp53range.stop()
+    });
   });
 });