Add data-output tests for plotPCA and plotFingerprint

pydeeptools/deeptools/test/test_plotFingerprint.py

@@ -17,7 +17,37 @@ def test_plotFingerprint_default():
     deeptools.plotFingerprint.main(args)
 
     res = compare_images(ROOT + 'test_plotFingerprint_default.png', plotfile.name, tolerance)
-    
+
     assert res is None, res
-
-    os.remove(plotfile.name)
+
+    os.remove(plotfile.name)
+
+
+def test_plotFingerprint_quality_metrics_and_JSD():
+    """
+    Test --outQualityMetrics together with --JSDsample. Neither the quality
+    metrics table nor the JS-distance computation was covered elsewhere.
+    """
+    plotfile = NamedTemporaryFile(suffix='.png', prefix='deeptools_testfile_', delete=False)
+    qcfile = NamedTemporaryFile(suffix='.tab', prefix='deeptools_testfile_', delete=False)
+    args = ("-b {0}test1.bam {0}test2.bam -o {1} --plotFileFormat png -l test1 test2 "
+            "--outQualityMetrics {2} --JSDsample {0}test1.bam".format(TEST_DATA, plotfile.name, qcfile.name)).split()
+    deeptools.plotFingerprint.main(args)
+
+    with open(qcfile.name) as _foo:
+        lines = [line.rstrip("\n").split("\t") for line in _foo]
+
+    # header + one row per sample
+    assert len(lines) == 3, f"expected 3 lines, got {len(lines)}"
+    header = lines[0]
+    auc = header.index("AUC")
+    jsd = header.index("JS Distance")
+
+    rows = {row[0]: row for row in lines[1:]}
+    assert abs(float(rows["test1"][auc]) - 0.39310288701202156) < 1e-4
+    assert abs(float(rows["test2"][auc]) - 0.3641251150405128) < 1e-4
+    # JS distance of the JSDsample (test1) against itself is nan; test2 is finite
+    assert abs(float(rows["test2"][jsd]) - 0.078613413909822) < 1e-4
+
+    os.remove(plotfile.name)
+    os.remove(qcfile.name)

pydeeptools/deeptools/test/test_plotPCA.py

@@ -1,5 +1,6 @@
 import os
 import filecmp
+import numpy as np
 from matplotlib.testing.compare import compare_images
 from tempfile import NamedTemporaryFile
 import deeptools.plotPCA
@@ -20,6 +21,34 @@ def test_plotPCA_default():
     res = compare_images(ROOT + 'test_plotPCA_default.png', plotfile.name, tolerance)
     assert res is None, res
     #assert filecmp.cmp(os.path.join(ROOT, 'test_plotPCA_default.tsv'), tsvfile.name) is True
-
+
+    os.remove(plotfile.name)
+    os.remove(tsvfile.name)
+
+
+def test_plotPCA_outFileNameData():
+    """
+    Verify the numeric --outFileNameData output. The eigenvector sign is
+    arbitrary (and can flip across BLAS/platforms), so we assert on the
+    sign-independent eigenvalue column and the table shape rather than the
+    projected coordinates.
+    """
+    plotfile = NamedTemporaryFile(suffix='.png', prefix='deeptools_testfile_', delete=False)
+    tsvfile = NamedTemporaryFile(suffix='.tsv', prefix='deeptools_testfile_', delete=False)
+    args = "-in {0}test_samples.npz -o {1} --outFileNameData {2}".format(TEST_DATA, plotfile.name, tsvfile.name).split()
+    deeptools.plotPCA.main(args)
+
+    # Columns: Component, wt1, wt2, wt3, kd1, kd2, kd3, Eigenvalue
+    data = np.loadtxt(tsvfile.name, skiprows=1)
+    assert data.shape == (6, 8), f"unexpected shape {data.shape}"
+    # Component index column
+    np.testing.assert_array_equal(data[:, 0], np.arange(1, 7))
+    eigenvalues = data[:, -1]
+    expected_eigenvalues = np.array([
+        5.807692278755936, 0.07423028883557825, 0.04897177773493676,
+        0.03680941552538939, 0.026706723301448194, 0.017613563942900697,
+    ])
+    np.testing.assert_allclose(eigenvalues, expected_eigenvalues, rtol=1e-5)
+
     os.remove(plotfile.name)
     os.remove(tsvfile.name)