add more complete unit tests

cmdoret · Oct 7, 2020 · 77f09db · 77f09db
1 parent ecc8308
commit 77f09db
Show file tree

Hide file tree

Showing 5 changed files with 25 additions and 5 deletions.
diff --git a/dnaglider/pkg/kmers.go b/dnaglider/pkg/kmers.go
@@ -55,6 +55,9 @@ func (p *KmerProfile) CountsToFreqs() {
 func (p *KmerProfile) KmerDist(ref KmerProfile) float64 {
 	var dist float64
 	for kmer, freq := range ref.Profile {
+		// This works because etching missing k-mer
+		// returns zero value for float
+
 		dist += math.Pow(freq-p.Profile[kmer], 2.0)
 	}
 	dist = math.Sqrt(dist)

diff --git a/dnaglider/pkg/stats.go b/dnaglider/pkg/stats.go
@@ -53,7 +53,7 @@ func SeqEntropy(seq *seq.Seq) float64 {
 	return -entro
 }
 
-// SeqKmer will compute the Kmer profile of the input profile
+// SeqKmerDiv will compute the Kmer profile of the input profile
 // and compute its distance to a reference k-mer profile.
 func SeqKmerDiv(seq *seq.Seq, ref KmerProfile) float64 {
 	// Get the k-mer length from field name, compute the

diff --git a/dnaglider/pkg/stats_test.go b/dnaglider/pkg/stats_test.go
@@ -66,3 +66,23 @@ func TestSeqEntropy(t *testing.T) {
 	}
 
 }
+
+func TestSeqKmerDiv(t *testing.T) {
+	refSeq, _ := seq.NewSeq(seq.DNA, []byte("AAATAA"))
+	querySeq, _ := seq.NewSeq(seq.DNA, []byte("AAA"))
+
+	refProf := NewKmerProfile(3)
+	refProf.GetSeqKmers(refSeq)
+	refProf.CountsToFreqs()
+	// ref:  AAA, AAT, ATA, TAA
+	// prof: AAA
+	expDist := math.Sqrt(0.75*0.75 + 3*(0.25*0.25))
+	obsDist := SeqKmerDiv(querySeq, refProf)
+	if expDist != obsDist {
+		t.Errorf(
+			"Wrong kmer distance with reference: got %f instead of %f",
+			obsDist,
+			expDist,
+		)
+	}
+}
diff --git a/dnaglider/pkg/windows.go b/dnaglider/pkg/windows.go
@@ -86,9 +86,6 @@ func ChunkGenome(records <-chan fastx.Record, winSize int, winStride int, chunkS
 					Seq:     rec.Seq.SubSeq(bpStart, bpEnd),
 				}
 
-				if len(chunk.Starts) == 0 {
-					fmt.Println("empty chunk")
-				}
 				chunks <- chunk
 				bpStart = bpEnd + 1
 			}

diff --git a/dnaglider/pkg/windows_test.go b/dnaglider/pkg/windows_test.go
@@ -49,7 +49,7 @@ func TestConsumeChunks(t *testing.T) {
 	chunks := ChunkGenome(records, WINSIZE, WINSTRIDE, CHUNKSIZE)
 	expHeader := []string{"chrom", "start", "end", "GC", "GCSKEW", "ENTRO", "3MER"}
 
-	for res := range ConsumeChunks(chunks, []string{"GC", "GCSKEW", "ENTRO"}, ref) {
+	for res := range ConsumeChunks(chunks, []string{"GC", "GCSKEW", "ENTRO", "3MER"}, ref) {
 		for i := range res.Header {
 			if res.Header[i] != expHeader[i] {
 				t.Errorf("Result header is incorrect.: %s", res.Header)