pepdata 1.3.5__tar.gz → 1.3.7__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (48) hide show
  1. {pepdata-1.3.5/pepdata.egg-info → pepdata-1.3.7}/PKG-INFO +1 -1
  2. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/BLOSUM50 +1 -1
  3. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/peptide_vectorizer.py +24 -11
  4. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/version.py +1 -1
  5. {pepdata-1.3.5 → pepdata-1.3.7/pepdata.egg-info}/PKG-INFO +1 -1
  6. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_blosum.py +12 -10
  7. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_ngram.py +35 -0
  8. {pepdata-1.3.5 → pepdata-1.3.7}/LICENSE.txt +0 -0
  9. {pepdata-1.3.5 → pepdata-1.3.7}/MANIFEST.in +0 -0
  10. {pepdata-1.3.5 → pepdata-1.3.7}/README.md +0 -0
  11. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/__init__.py +0 -0
  12. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/amino_acid.py +0 -0
  13. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/amino_acid_alphabet.py +0 -0
  14. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/amino_acid_properties.py +0 -0
  15. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/blosum.py +0 -0
  16. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/chou_fasman.py +0 -0
  17. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/common.py +0 -0
  18. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/iedb/__init__.py +0 -0
  19. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/iedb/alleles.py +0 -0
  20. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/iedb/columns.py +0 -0
  21. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/iedb/common.py +0 -0
  22. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/iedb/mhc.py +0 -0
  23. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/iedb/tcell.py +0 -0
  24. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/BLOSUM30 +0 -0
  25. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/BLOSUM62 +0 -0
  26. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/__init__.py +0 -0
  27. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/amino_acid_properties.txt +0 -0
  28. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/helix_vs_coil.txt +0 -0
  29. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/helix_vs_strand.txt +0 -0
  30. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/pmbec.mat +0 -0
  31. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/matrices/strand_vs_coil.txt +0 -0
  32. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/pmbec.py +0 -0
  33. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/reduced_alphabet.py +0 -0
  34. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/residue_contact_energies.py +0 -0
  35. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata/static_data.py +0 -0
  36. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata.egg-info/SOURCES.txt +0 -0
  37. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata.egg-info/dependency_links.txt +0 -0
  38. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata.egg-info/requires.txt +0 -0
  39. {pepdata-1.3.5 → pepdata-1.3.7}/pepdata.egg-info/top_level.txt +0 -0
  40. {pepdata-1.3.5 → pepdata-1.3.7}/pyproject.toml +0 -0
  41. {pepdata-1.3.5 → pepdata-1.3.7}/requirements.txt +0 -0
  42. {pepdata-1.3.5 → pepdata-1.3.7}/setup.cfg +0 -0
  43. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_amino_acids.py +0 -0
  44. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_iedb_alleles.py +0 -0
  45. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_iedb_cache_lifecycle.py +0 -0
  46. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_iedb_mhc.py +0 -0
  47. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_iedb_tcell.py +0 -0
  48. {pepdata-1.3.5 → pepdata-1.3.7}/tests/test_pmbec.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: pepdata
3
- Version: 1.3.5
3
+ Version: 1.3.7
4
4
  Summary: Immunological peptide datasets and amino acid properties
5
5
  Author-email: Alex Rubinsteyn <alex.rubinsteyn@mssm.edu>
6
6
  Project-URL: Homepage, https://github.com/openvax/pepdata
@@ -16,6 +16,6 @@ F -3 -3 -4 -5 -2 -4 -3 -4 -1 0 1 -4 0 8 -4 -3 -2 1 4 -1
16
16
  P -1 -3 -2 -1 -4 -1 -1 -2 -2 -3 -4 -1 -3 -4 10 -1 -1 -4 -3 -3
17
17
  S 1 -1 1 0 -1 0 -1 0 -1 -3 -3 0 -2 -3 -1 5 2 -4 -2 -2
18
18
  T 0 -1 0 -1 -1 -1 -1 -2 -2 -1 -1 -1 -1 -2 -1 2 5 -3 -2 0
19
- W -3 -3 -4 -5 -5 -1 -3 -3 -3 -3 -2 -3 -1 1 -4 -4 -4 15 2 -3
19
+ W -3 -3 -4 -5 -5 -1 -3 -3 -3 -3 -2 -3 -1 1 -4 -4 -3 15 2 -3
20
20
  Y -2 -1 -2 -3 -3 -1 -2 -3 2 -1 -1 -2 0 4 -3 -2 -2 2 8 -1
21
21
  V 0 -3 -3 -4 -1 -3 -3 -4 -4 4 1 -3 1 -1 -3 -2 0 -3 -1 5
@@ -13,6 +13,8 @@
13
13
  # limitations under the License.
14
14
 
15
15
 
16
+ from copy import copy
17
+
16
18
  import numpy as np
17
19
  from sklearn.feature_extraction.text import CountVectorizer
18
20
  from sklearn.preprocessing import normalize
@@ -70,16 +72,12 @@ class PeptideVectorizer(object):
70
72
  n-grams) with n-gram counts, or per-row frequencies (each row
71
73
  divided by its total) when normalize_row is True.
72
74
  """
73
- self.count_vectorizer = \
74
- make_count_vectorizer(self.reduced_alphabet, self.max_ngram)
75
-
76
- if self.training_already_reduced:
77
- c = make_count_vectorizer(None, self.max_ngram)
78
- X = c.fit_transform(amino_acid_strings).toarray()
79
- self.count_vectorizer.vocabulary_ = c.vocabulary_
80
- else:
81
- c = self.count_vectorizer
82
- X = c.fit_transform(amino_acid_strings).toarray()
75
+ # Already-reduced training must use the same tokenization later used
76
+ # for reduced transform input. Keep its existing lowercase vocabulary
77
+ # and feature-column order, including in saved vectorizers.
78
+ alphabet = None if self.training_already_reduced else self.reduced_alphabet
79
+ self.count_vectorizer = make_count_vectorizer(alphabet, self.max_ngram)
80
+ X = self.count_vectorizer.fit_transform(amino_acid_strings).toarray()
83
81
 
84
82
  if self.normalize_row:
85
83
  X = normalize(X, norm='l1')
@@ -93,6 +91,8 @@ class PeptideVectorizer(object):
93
91
  Vectorize sequences with the vocabulary learned by fit.
94
92
 
95
93
  n-grams that weren't seen during fitting are not counted.
94
+ Input sequences are unreduced, even when training_already_reduced was
95
+ used during fitting.
96
96
 
97
97
  Parameters
98
98
  ----------
@@ -106,7 +106,20 @@ class PeptideVectorizer(object):
106
106
  vocabulary), normalized per row as in fit_transform.
107
107
  """
108
108
  assert self.count_vectorizer, "Must call 'fit' before 'transform'"
109
- X = self.count_vectorizer.transform(amino_acid_strings).toarray()
109
+ vectorizer = self.count_vectorizer
110
+ if self.training_already_reduced:
111
+ if self.reduced_alphabet is not None:
112
+ amino_acid_strings = [
113
+ "".join(self.reduced_alphabet[residue] for residue in sequence)
114
+ for sequence in amino_acid_strings
115
+ ]
116
+ # Older saved models attached the raw-sequence preprocessor to a
117
+ # lowercase vocabulary fitted by a different CountVectorizer.
118
+ # Preserve that vocabulary without reducing twice or mutating it.
119
+ if vectorizer.preprocessor is not None:
120
+ vectorizer = copy(vectorizer)
121
+ vectorizer.preprocessor = None
122
+ X = vectorizer.transform(amino_acid_strings).toarray()
110
123
  if self.normalize_row:
111
124
  X = normalize(X, norm='l1')
112
125
  return X
@@ -1,4 +1,4 @@
1
- __version__ = "1.3.5"
1
+ __version__ = "1.3.7"
2
2
 
3
3
 
4
4
  def print_version():
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: pepdata
3
- Version: 1.3.5
3
+ Version: 1.3.7
4
4
  Summary: Immunological peptide datasets and amino acid properties
5
5
  Author-email: Alex Rubinsteyn <alex.rubinsteyn@mssm.edu>
6
6
  Project-URL: Homepage, https://github.com/openvax/pepdata
@@ -1,3 +1,5 @@
1
+ from pathlib import Path
2
+
1
3
  import pytest
2
4
 
3
5
  from pepdata.amino_acid_alphabet import canonical_amino_acid_letters
@@ -41,7 +43,7 @@ def test_blosum_diagonal_scores(number):
41
43
  assert table[residue][residue] == max(row)
42
44
 
43
45
 
44
- @pytest.mark.parametrize("number", [30, 62])
46
+ @pytest.mark.parametrize("number", sorted(BLOSUM_TABLES))
45
47
  def test_blosum_is_symmetric(number):
46
48
  table, _ = BLOSUM_TABLES[number]
47
49
  for x in canonical_amino_acid_letters:
@@ -49,15 +51,15 @@ def test_blosum_is_symmetric(number):
49
51
  assert table[x][y] == table[y][x]
50
52
 
51
53
 
52
- def test_blosum50_asymmetry_is_limited_to_shipped_w_t_entry():
53
- # The shipped BLOSUM50 file has W->T = -4 but T->W = -3 (published value -3).
54
- asymmetric = {
55
- (x, y)
56
- for x in canonical_amino_acid_letters
57
- for y in canonical_amino_acid_letters
58
- if blosum50_dict[x][y] != blosum50_dict[y][x]
59
- }
60
- assert asymmetric == {("T", "W"), ("W", "T")}
54
+ def test_blosum50_matches_ncbi_canonical_scores():
55
+ reference = Path(__file__).parent / "data/ncbi_BLOSUM50.txt"
56
+ rows = [line.split() for line in reference.read_text().splitlines()
57
+ if line.strip() and not line.startswith("#")]
58
+ header = rows[0]
59
+ expected = {row[0]: dict(zip(header, map(int, row[1:]))) for row in rows[1:]}
60
+ for x in canonical_amino_acid_letters:
61
+ for y in canonical_amino_acid_letters:
62
+ assert blosum50_dict[x][y] == expected[x][y], (x, y)
61
63
 
62
64
 
63
65
  @pytest.mark.parametrize("number", sorted(BLOSUM_TABLES))
@@ -18,6 +18,7 @@ import numpy as np
18
18
  import pytest
19
19
 
20
20
  from pepdata import PeptideVectorizer
21
+ from pepdata.peptide_vectorizer import make_count_vectorizer
21
22
  from pepdata.reduced_alphabet import hp2
22
23
 
23
24
  # isoforms of two different proteins a, b
@@ -117,3 +118,37 @@ def test_pickled_vectorizer_transforms_identically():
117
118
  vectorizer.fit(A)
118
119
  restored = pickle.loads(pickle.dumps(vectorizer))
119
120
  np.testing.assert_allclose(restored.transform(B), vectorizer.transform(B))
121
+
122
+
123
+ @pytest.mark.parametrize("max_ngram", [1, 2])
124
+ @pytest.mark.parametrize("normalize_row", [False, True])
125
+ def test_already_reduced_training_matches_raw_transform(max_ngram, normalize_row):
126
+ sequences = ["SIINFEKL", "GILGFVFTL"]
127
+ reduced = ["".join(hp2[r] for r in sequence) for sequence in sequences]
128
+ vectorizer = PeptideVectorizer(
129
+ max_ngram=max_ngram, reduced_alphabet=hp2,
130
+ training_already_reduced=True, normalize_row=normalize_row)
131
+ expected = vectorizer.fit_transform(reduced)
132
+ assert expected.sum() > 0
133
+ vocabulary = dict(vectorizer.count_vectorizer.vocabulary_)
134
+ np.testing.assert_allclose(vectorizer.transform(sequences), expected)
135
+ restored = pickle.loads(pickle.dumps(vectorizer))
136
+ np.testing.assert_allclose(restored.transform(sequences), expected)
137
+ assert restored.count_vectorizer.vocabulary_ == vocabulary
138
+
139
+
140
+ def test_legacy_reduced_vectorizer_preserves_lowercase_feature_columns():
141
+ sequences = ["SIINFEKL", "GILGFVFTL"]
142
+ reduced = ["".join(hp2[r] for r in sequence) for sequence in sequences]
143
+ fitted = make_count_vectorizer(None, 2)
144
+ expected = fitted.fit_transform(reduced).toarray()
145
+ legacy = PeptideVectorizer(
146
+ max_ngram=2, reduced_alphabet=hp2,
147
+ training_already_reduced=True, normalize_row=False)
148
+ legacy.count_vectorizer = make_count_vectorizer(hp2, 2)
149
+ legacy.count_vectorizer.vocabulary_ = fitted.vocabulary_
150
+ original_preprocessor = legacy.count_vectorizer.preprocessor
151
+ vocabulary = dict(fitted.vocabulary_)
152
+ np.testing.assert_array_equal(legacy.transform(sequences), expected)
153
+ assert legacy.count_vectorizer.preprocessor is original_preprocessor
154
+ assert legacy.count_vectorizer.vocabulary_ == vocabulary
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes
File without changes