Merge pull request #87 from MaozGelbart/motifpssm_fixes

FIX: MotifPssmPattern load from file now loads correctly

dnachisel/SequencePattern/MotifPssmPattern.py

@@ -150,7 +150,7 @@ def list_from_file(
           sequence(s) with the absolute highest possible score".
         """
         if isinstance(motifs_file, str):
-            with open("./jaspar.txt", "r") as f:
+            with open(motifs_file, "r") as f:
                 motifs_list = motifs.parse(f, file_format)
         else:
             motifs_list = motifs.parse(motifs_file, file_format)

tests/data/multiple_motifs.meme.txt

@@ -0,0 +1,33 @@
+MEME version 4
+
+ALPHABET= ACGT
+
+strands: + -
+
+Background letter frequencies
+A 0.25 C 0.25 G 0.25 T 0.25
+
+MOTIF MA0016.1 MA0016.1.usp
+letter-probability matrix: alength= 4 w= 10 nsites= 38 E= 0
+ 0.000000  0.026316  0.973684  0.000000
+ 0.026316  0.000000  0.947368  0.026316
+ 0.000000  0.000000  1.000000  0.000000
+ 0.000000  0.000000  1.000000  0.000000
+ 0.000000  0.000000  0.000000  1.000000
+ 0.000000  0.947368  0.026316  0.026316
+ 0.921053  0.000000  0.078947  0.000000
+ 0.131579  0.657895  0.078947  0.131579
+ 0.131579  0.210526  0.578947  0.078947
+ 0.157895  0.263158  0.421053  0.157895
+URL http://jaspar.genereg.net/matrix/MA0016.1
+
+MOTIF MA0011.2 MA0011.2.br
+letter-probability matrix: alength= 4 w= 6 nsites= 12 E= 0
+ 0.000000  0.833333  0.000000  0.166667
+ 0.000000  0.083333  0.000000  0.916667
+ 1.000000  0.000000  0.000000  0.000000
+ 0.083333  0.083333  0.166667  0.666667
+ 0.166667  0.000000  0.083333  0.750000
+ 0.083333  0.166667  0.083333  0.666667
+URL http://jaspar.genereg.net/matrix/MA0011.2
+

tests/data/single_motif.meme.txt

@@ -0,0 +1,19 @@
+MEME version 4
+
+ALPHABET= ACGT
+
+strands: + -
+
+Background letter frequencies
+A 0.25 C 0.25 G 0.25 T 0.25
+
+MOTIF MA0011.2 MA0011.2.br
+letter-probability matrix: alength= 4 w= 6 nsites= 12 E= 0
+ 0.000000  0.833333  0.000000  0.166667
+ 0.000000  0.083333  0.000000  0.916667
+ 1.000000  0.000000  0.000000  0.000000
+ 0.083333  0.083333  0.166667  0.666667
+ 0.166667  0.000000  0.083333  0.750000
+ 0.083333  0.166667  0.083333  0.666667
+URL http://jaspar.genereg.net/matrix/MA0011.2
+

tests/test_patterns.py

@@ -1,9 +1,39 @@
-from dnachisel.SequencePattern import SequencePattern
+import pytest
+from pathlib import Path
+
+from dnachisel import SequencePattern, MotifPssmPattern
+
+
+@pytest.fixture
+def test_single_motif_filepath():
+    return str(Path(__file__).parent / 'data' / 'single_motif.meme.txt')
+
+
+@pytest.fixture
+def test_multiple_motif_filepath():
+    return str(Path(__file__).parent / 'data' / 'multiple_motifs.meme.txt')
+
 
 def test_patterns_from_string():
     pattern = SequencePattern.from_string("6xT")
     assert pattern.expression == "TTTTTT"
     pattern = SequencePattern.from_string("BsmBI_site")
     assert pattern.expression == "CGTCTC"
     pattern = SequencePattern.from_string("5x2mer")
-    assert pattern.expression == '([ATGC]{2})\\1{4}'
+    assert pattern.expression == '([ATGC]{2})\\1{4}'
+
+
+def test_pssm_pattern_from_file(
+    test_single_motif_filepath, test_multiple_motif_filepath
+):
+    single_pattern = MotifPssmPattern.list_from_file(
+        test_single_motif_filepath, "minimal", relative_threshold=0.9
+    )
+    assert len(single_pattern) == 1
+    assert all([isinstance(p, MotifPssmPattern) for p in single_pattern])
+
+    multiple_patterns = MotifPssmPattern.list_from_file(
+        test_multiple_motif_filepath, "minimal", relative_threshold=0.9
+    )
+    assert len(multiple_patterns) == 2
+    assert all([isinstance(p, MotifPssmPattern) for p in multiple_patterns])