RFCT Simplify norm_abundance function

SemiBin/utils.py

@@ -659,12 +659,7 @@ def compute_min_length(min_length, fafile, ratio):
 def norm_abundance(data):
     import numpy as np
     n = data.shape[1] - 136
-    flag = False
 
     if n >= 20:
-        flag = True
-    else:
-        if n >= 5:
-            if np.mean(np.sum(data[:, 136:], axis=1)) > 2:
-                flag = True
-    return flag
+        return True
+    return (n >= 5 and np.mean(np.sum(data[:, 136:], axis=1)) > 2)

test/test_utils.py

@@ -1,4 +1,4 @@
-from SemiBin.utils import get_must_link_threshold, get_marker, split_data, n50_l50, extract_bams
+from SemiBin.utils import get_must_link_threshold, get_marker, split_data, n50_l50, extract_bams, norm_abundance
 from hypothesis import given, strategies as st
 from io import StringIO
 import numpy as np
@@ -135,3 +135,16 @@ def test_extract_bams(tmpdir):
              tmpdir)
     assert rs is None
 
+def test_norm_abundance():
+    assert not norm_abundance(np.random.randn(10, 136))
+    assert not norm_abundance(np.random.randn(12, 137))
+    assert not norm_abundance(np.random.randn(12, 140))
+    assert not norm_abundance(np.random.randn(12, 148))
+    assert not norm_abundance(np.abs(np.random.randn(12, 138))*2)
+    assert not norm_abundance(np.abs(np.random.randn(12, 138))*4)
+
+    assert norm_abundance(np.abs(np.random.randn(12, 148)))
+    assert norm_abundance(       np.random.randn(12, 156) )
+    assert norm_abundance(       np.random.randn(12, 164) )
+    assert norm_abundance(np.abs(np.random.randn(12, 164)))
+