[ADAM-1728] Expose region join API through Python and R.

Resolves #1728.

adam-python/bdgenomics/adam/test/alignmentRecordRdd_test.py

@@ -236,3 +236,123 @@ def test_persisting(self):
         persistedReads.unpersist()
         cached = self.sc._jsc.getPersistentRDDs()
         self.assertEquals(cached.isEmpty(), True)
+
+
+    def test_broadcast_inner_join(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.broadcastRegionJoin(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 5)
+
+
+    def test_broadcast_right_outer_join(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.rightOuterBroadcastRegionJoin(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 6)
+
+
+    def test_shuffle_inner_join(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.shuffleRegionJoin(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 5)
+
+
+    def test_shuffle_right_outer_join(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.rightOuterShuffleRegionJoin(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 6)
+
+
+    def test_shuffle_left_outer_join(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.leftOuterShuffleRegionJoin(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 20)
+
+
+    def test_shuffle_full_outer_join(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.fullOuterShuffleRegionJoin(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 21)
+
+
+    def test_shuffle_inner_join_groupBy_left(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.shuffleRegionJoinAndGroupByLeft(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 5)
+
+
+    def test_shuffle_right_outer_join_groupBy_left(self):
+
+        readsPath = self.resourceFile("small.1.sam")
+        targetsPath = self.resourceFile("small.1.bed")
+
+        ac = ADAMContext(self.ss)
+
+        reads = ac.loadAlignments(readsPath)
+        targets = ac.loadFeatures(targetsPath)
+
+        jRdd = reads.rightOuterShuffleRegionJoinAndGroupByLeft(targets)
+
+        self.assertEquals(jRdd.toDF().count(), 21)

adam-r/bdgenomics.adam/NAMESPACE

@@ -2,13 +2,18 @@
 
 export(ADAMContext)
 export(aggregatedCoverage)
+export(broadcastRegionJoin)
+export(broadcastRegionJoinAndGroupByRight)
 export(cache)
 export(collapse)
 export(countKmers)
 export(coverage)
 export(createADAMContext)
 export(flankAdjacentFragments)
 export(flatten)
+export(fullOuterShuffleRegionJoin)
+export(leftOuterShuffleRegionJoin)
+export(leftOuterShuffleRegionJoinAndGroupByLeft)
 export(loadAlignments)
 export(loadContigFragments)
 export(loadCoverage)
@@ -21,10 +26,16 @@ export(persist)
 export(pipe)
 export(realignIndels)
 export(recalibrateBaseQualities)
+export(rightOuterBroadcastRegionJoin)
+export(rightOuterBroadcastRegionJoinAndGroupByRight)
+export(rightOuterShuffleRegionJoin)
+export(rightOuterShuffleRegionJoinAndGroupByLeft)
 export(save)
 export(saveAsParquet)
 export(saveAsSam)
 export(saveAsVcf)
+export(shuffleRegionJoin)
+export(shuffleRegionJoinAndGroupByLeft)
 export(sort)
 export(sortLexicographically)
 export(sortReadsByReferencePosition)
@@ -45,18 +56,22 @@ exportClasses(CoverageRDD)
 exportClasses(FeatureRDD)
 exportClasses(FragmentRDD)
 exportClasses(GenomicDataset)
-exportClasses(GenomicRDD)
 exportClasses(GenotypeRDD)
 exportClasses(NucleotideContigFragmentRDD)
 exportClasses(VariantContextRDD)
 exportClasses(VariantRDD)
 exportMethods(aggregatedCoverage)
+exportMethods(broadcastRegionJoin)
+exportMethods(broadcastRegionJoinAndGroupByRight)
 exportMethods(cache)
 exportMethods(collapse)
 exportMethods(countKmers)
 exportMethods(coverage)
 exportMethods(flankAdjacentFragments)
 exportMethods(flatten)
+exportMethods(fullOuterShuffleRegionJoin)
+exportMethods(leftOuterShuffleRegionJoin)
+exportMethods(leftOuterShuffleRegionJoinAndGroupByLeft)
 exportMethods(loadAlignments)
 exportMethods(loadContigFragments)
 exportMethods(loadCoverage)
@@ -69,10 +84,16 @@ exportMethods(persist)
 exportMethods(pipe)
 exportMethods(realignIndels)
 exportMethods(recalibrateBaseQualities)
+exportMethods(rightOuterBroadcastRegionJoin)
+exportMethods(rightOuterBroadcastRegionJoinAndGroupByRight)
+exportMethods(rightOuterShuffleRegionJoin)
+exportMethods(rightOuterShuffleRegionJoinAndGroupByLeft)
 exportMethods(save)
 exportMethods(saveAsParquet)
 exportMethods(saveAsSam)
 exportMethods(saveAsVcf)
+exportMethods(shuffleRegionJoin)
+exportMethods(shuffleRegionJoinAndGroupByLeft)
 exportMethods(sort)
 exportMethods(sortLexicographically)
 exportMethods(sortReadsByReferencePosition)