X-Git-Url: https://git.arvados.org/lightning.git/blobdiff_plain/d81fc575f556e71b5c552d03c626b43c0744b45f..5d939b4ccdab73d0729db56219ba674fb9995c26:/slice_test.go

diff --git a/slice_test.go b/slice_test.go
index 7332100634..47c43e1cee 100644
--- a/slice_test.go
+++ b/slice_test.go
@@ -81,6 +81,24 @@ func (s *sliceSuite) TestImportAndSlice(c *check.C) {
 	out, _ := exec.Command("find", slicedir, "-ls").CombinedOutput()
 	c.Logf("%s", out)
 
+	c.Log("=== dump ===")
+	{
+		dumpdir := c.MkDir()
+		exited = (&dump{}).RunCommand("dump", []string{
+			"-local=true",
+			"-tags=4,6,7",
+			"-input-dir=" + slicedir,
+			"-output-dir=" + dumpdir,
+		}, nil, os.Stderr, os.Stderr)
+		c.Check(exited, check.Equals, 0)
+		out, _ := exec.Command("find", dumpdir, "-ls").CombinedOutput()
+		c.Logf("%s", out)
+		dumped, err := ioutil.ReadFile(dumpdir + "/variants.csv")
+		c.Assert(err, check.IsNil)
+		c.Logf("%s", dumped)
+		c.Check(string(dumped), check.Matches, `(?ms).*\n6,1,1,chr2,349,AAAACTG.*`)
+	}
+
 	c.Log("=== slice-numpy ===")
 	{
 		npydir := c.MkDir()
@@ -134,6 +152,7 @@ func (s *sliceSuite) TestImportAndSlice(c *check.C) {
 			"-regions=" + tmpdir + "/chr1-12-100.bed",
 			"-input-dir=" + slicedir,
 			"-output-dir=" + npydir,
+			"-chunked-hgvs-matrix=true",
 		}, nil, os.Stderr, os.Stderr)
 		c.Check(exited, check.Equals, 0)
 		out, _ := exec.Command("find", npydir, "-ls").CombinedOutput()
@@ -160,9 +179,172 @@ func (s *sliceSuite) TestImportAndSlice(c *check.C) {
 			c.Check(string(annotations), check.Matches, "(?ms).*"+s+".*")
 		}
 
-		annotations, err = ioutil.ReadFile(npydir + "/matrix.0002.annotations.csv")
+		for _, fnm := range []string{
+			npydir + "/matrix.0001.annotations.csv",
+			npydir + "/matrix.0002.annotations.csv",
+		} {
+			annotations, err := ioutil.ReadFile(fnm)
+			c.Assert(err, check.IsNil)
+			c.Check(string(annotations), check.Equals, "", check.Commentf(fnm))
+		}
+	}
+
+	err = ioutil.WriteFile(tmpdir+"/chr1and2-100-200.bed", []byte("chr1\t100\t200\ttest.1\nchr2\t100\t200\ttest.2\n"), 0644)
+	c.Check(err, check.IsNil)
+
+	c.Log("=== slice-numpy + regions + merge ===")
+	{
+		npydir := c.MkDir()
+		exited := (&sliceNumpy{}).RunCommand("slice-numpy", []string{
+			"-local=true",
+			"-regions=" + tmpdir + "/chr1and2-100-200.bed",
+			"-input-dir=" + slicedir,
+			"-output-dir=" + npydir,
+			"-merge-output=true",
+			"-single-hgvs-matrix=true",
+		}, nil, os.Stderr, os.Stderr)
+		c.Check(exited, check.Equals, 0)
+		out, _ := exec.Command("find", npydir, "-ls").CombinedOutput()
+		c.Logf("%s", out)
+
+		f, err := os.Open(npydir + "/matrix.npy")
+		c.Assert(err, check.IsNil)
+		defer f.Close()
+		npy, err := gonpy.NewReader(f)
+		c.Assert(err, check.IsNil)
+		c.Check(npy.Shape, check.DeepEquals, []int{4, 4})
+		variants, err := npy.GetInt16()
+		if c.Check(err, check.IsNil) {
+			c.Check(variants, check.DeepEquals, []int16{2, 1, 3, 1, -1, -1, 4, 2, 2, 1, 3, 1, -1, -1, 4, 2})
+		}
+
+		annotations, err := ioutil.ReadFile(npydir + "/matrix.annotations.csv")
 		c.Assert(err, check.IsNil)
 		c.Logf("%s", annotations)
-		c.Check(string(annotations), check.Equals, "")
+		for _, s := range []string{
+			"0,0,1,chr1:g.161A>T",
+			"0,0,1,chr1:g.178A>T",
+			"4,1,2,chr2:g.125_127delinsAAA",
+		} {
+			c.Check(string(annotations), check.Matches, "(?ms).*"+s+".*")
+		}
+	}
+
+	c.Log("=== slice-numpy + chunked hgvs matrix ===")
+	{
+		err = ioutil.WriteFile(tmpdir+"/casecontrol.tsv", []byte(`SampleID	CC
+pipeline1/input1	1
+pipeline1/input2	0
+pipeline1dup/input1	1
+pipeline1dup/input2	0
+`), 0600)
+		c.Assert(err, check.IsNil)
+		npydir := c.MkDir()
+		exited := (&sliceNumpy{}).RunCommand("slice-numpy", []string{
+			"-local=true",
+			"-chunked-hgvs-matrix=true",
+			"-chi2-case-control-file=" + tmpdir + "/casecontrol.tsv",
+			"-chi2-case-control-column=CC",
+			"-chi2-p-value=0.5",
+			"-min-coverage=0.75",
+			"-input-dir=" + slicedir,
+			"-output-dir=" + npydir,
+		}, nil, os.Stderr, os.Stderr)
+		c.Check(exited, check.Equals, 0)
+		out, _ := exec.Command("find", npydir, "-ls").CombinedOutput()
+		c.Logf("%s", out)
+
+		annotations, err := ioutil.ReadFile(npydir + "/hgvs.chr2.annotations.csv")
+		c.Assert(err, check.IsNil)
+		c.Check(string(annotations), check.Equals, `0,chr2:g.470_472del
+1,chr2:g.471G>A
+2,chr2:g.472G>A
+`)
+	}
+
+	c.Log("=== slice-numpy + onehotChunked ===")
+	{
+		err = ioutil.WriteFile(tmpdir+"/casecontrol.tsv", []byte(`SampleID	CC
+pipeline1/input1	1
+pipeline1/input2	0
+pipeline1dup/input1	1
+pipeline1dup/input2	0
+`), 0600)
+		c.Assert(err, check.IsNil)
+		npydir := c.MkDir()
+		exited := (&sliceNumpy{}).RunCommand("slice-numpy", []string{
+			"-local=true",
+			"-chunked-onehot=true",
+			"-chi2-case-control-file=" + tmpdir + "/casecontrol.tsv",
+			"-chi2-case-control-column=CC",
+			"-chi2-p-value=0.5",
+			"-min-coverage=0.75",
+			"-input-dir=" + slicedir,
+			"-output-dir=" + npydir,
+		}, nil, os.Stderr, os.Stderr)
+		c.Check(exited, check.Equals, 0)
+		out, _ := exec.Command("find", npydir, "-ls").CombinedOutput()
+		c.Logf("%s", out)
+
+		f, err := os.Open(npydir + "/onehot.0002.npy")
+		c.Assert(err, check.IsNil)
+		defer f.Close()
+		npy, err := gonpy.NewReader(f)
+		c.Assert(err, check.IsNil)
+		c.Check(npy.Shape, check.DeepEquals, []int{4, 6})
+		onehot, err := npy.GetInt8()
+		if c.Check(err, check.IsNil) {
+			for r := 0; r < npy.Shape[0]; r++ {
+				c.Logf("%v", onehot[r*npy.Shape[1]:(r+1)*npy.Shape[1]])
+			}
+			c.Check(onehot, check.DeepEquals, []int8{
+				0, 0, 0, 1, 0, 0, // input1
+				0, 1, 0, 0, 0, 1, // input2
+				0, 0, 0, 1, 0, 0, // dup/input1
+				0, 1, 0, 0, 0, 1, // dup/input2
+			})
+		}
+	}
+
+	c.Log("=== slice-numpy + onehotSingle ===")
+	{
+		err = ioutil.WriteFile(tmpdir+"/casecontrol.tsv", []byte(`SampleID	CC
+pipeline1/input1	1
+pipeline1/input2	0
+pipeline1dup/input1	1
+pipeline1dup/input2	0
+`), 0600)
+		c.Assert(err, check.IsNil)
+		npydir := c.MkDir()
+		exited := (&sliceNumpy{}).RunCommand("slice-numpy", []string{
+			"-local=true",
+			"-single-onehot=true",
+			"-chi2-case-control-file=" + tmpdir + "/casecontrol.tsv",
+			"-chi2-case-control-column=CC",
+			"-chi2-p-value=0.5",
+			"-min-coverage=0.75",
+			"-input-dir=" + slicedir,
+			"-output-dir=" + npydir,
+		}, nil, os.Stderr, os.Stderr)
+		c.Check(exited, check.Equals, 0)
+		out, _ := exec.Command("find", npydir, "-ls").CombinedOutput()
+		c.Logf("%s", out)
+
+		f, err := os.Open(npydir + "/onehot.npy")
+		c.Assert(err, check.IsNil)
+		defer f.Close()
+		npy, err := gonpy.NewReader(f)
+		c.Assert(err, check.IsNil)
+		c.Check(npy.Shape, check.DeepEquals, []int{2, 16})
+		onehot, err := npy.GetUint32()
+		if c.Check(err, check.IsNil) {
+			for r := 0; r < npy.Shape[0]; r++ {
+				c.Logf("%v", onehot[r*npy.Shape[1]:(r+1)*npy.Shape[1]])
+			}
+			c.Check(onehot, check.DeepEquals, []uint32{
+				0, 2, 1, 3, 0, 2, 1, 3, 0, 2, 1, 3, 0, 2, 0, 2,
+				1, 1, 2, 2, 5, 5, 7, 7, 9, 9, 11, 11, 13, 13, 15, 15,
+			})
+		}
 	}
 }