annotate cpt_starts/gff3.py @ 3:e4c2da570be1 draft default tip

Uploaded
author cpt
date Fri, 20 May 2022 09:05:12 +0000
parents 9f2517655a1e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9f2517655a1e Uploaded
cpt
parents:
diff changeset
1 import copy
9f2517655a1e Uploaded
cpt
parents:
diff changeset
2 import logging
9f2517655a1e Uploaded
cpt
parents:
diff changeset
3
9f2517655a1e Uploaded
cpt
parents:
diff changeset
4 log = logging.getLogger()
9f2517655a1e Uploaded
cpt
parents:
diff changeset
5 log.setLevel(logging.WARN)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
6
9f2517655a1e Uploaded
cpt
parents:
diff changeset
7
9f2517655a1e Uploaded
cpt
parents:
diff changeset
8 def feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
9 feature_list,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
10 test,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
11 test_kwargs,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
12 subfeatures=True,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
13 parent=None,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
14 invert=False,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
15 recurse=True,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
16 ):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
17 """Recursively search through features, testing each with a test function, yielding matches.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
18
9f2517655a1e Uploaded
cpt
parents:
diff changeset
19 GFF3 is a hierachical data structure, so we need to be able to recursively
9f2517655a1e Uploaded
cpt
parents:
diff changeset
20 search through features. E.g. if you're looking for a feature with
9f2517655a1e Uploaded
cpt
parents:
diff changeset
21 ID='bob.42', you can't just do a simple list comprehension with a test
9f2517655a1e Uploaded
cpt
parents:
diff changeset
22 case. You don't know how deeply burried bob.42 will be in the feature tree. This is where feature_lambda steps in.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
23
9f2517655a1e Uploaded
cpt
parents:
diff changeset
24 :type feature_list: list
9f2517655a1e Uploaded
cpt
parents:
diff changeset
25 :param feature_list: an iterable of features
9f2517655a1e Uploaded
cpt
parents:
diff changeset
26
9f2517655a1e Uploaded
cpt
parents:
diff changeset
27 :type test: function reference
9f2517655a1e Uploaded
cpt
parents:
diff changeset
28 :param test: a closure with the method signature (feature, **kwargs) where
9f2517655a1e Uploaded
cpt
parents:
diff changeset
29 the kwargs are those passed in the next argument. This
9f2517655a1e Uploaded
cpt
parents:
diff changeset
30 function should return True or False, True if the feature is
9f2517655a1e Uploaded
cpt
parents:
diff changeset
31 to be yielded as part of the main feature_lambda function, or
9f2517655a1e Uploaded
cpt
parents:
diff changeset
32 False if it is to be ignored. This function CAN mutate the
9f2517655a1e Uploaded
cpt
parents:
diff changeset
33 features passed to it (think "apply").
9f2517655a1e Uploaded
cpt
parents:
diff changeset
34
9f2517655a1e Uploaded
cpt
parents:
diff changeset
35 :type test_kwargs: dictionary
9f2517655a1e Uploaded
cpt
parents:
diff changeset
36 :param test_kwargs: kwargs to pass to your closure when it is called.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
37
9f2517655a1e Uploaded
cpt
parents:
diff changeset
38 :type subfeatures: boolean
9f2517655a1e Uploaded
cpt
parents:
diff changeset
39 :param subfeatures: when a feature is matched, should just that feature be
9f2517655a1e Uploaded
cpt
parents:
diff changeset
40 yielded to the caller, or should the entire sub_feature
9f2517655a1e Uploaded
cpt
parents:
diff changeset
41 tree for that feature be included? subfeatures=True is
9f2517655a1e Uploaded
cpt
parents:
diff changeset
42 useful in cases such as searching for a gene feature,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
43 and wanting to know what RBS/Shine_Dalgarno_sequences
9f2517655a1e Uploaded
cpt
parents:
diff changeset
44 are in the sub_feature tree (which can be accomplished
9f2517655a1e Uploaded
cpt
parents:
diff changeset
45 with two feature_lambda calls). subfeatures=False is
9f2517655a1e Uploaded
cpt
parents:
diff changeset
46 useful in cases when you want to process (and possibly
9f2517655a1e Uploaded
cpt
parents:
diff changeset
47 return) the entire feature tree, such as applying a
9f2517655a1e Uploaded
cpt
parents:
diff changeset
48 qualifier to every single feature.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
49
9f2517655a1e Uploaded
cpt
parents:
diff changeset
50 :type invert: boolean
9f2517655a1e Uploaded
cpt
parents:
diff changeset
51 :param invert: Negate/invert the result of the filter.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
52
9f2517655a1e Uploaded
cpt
parents:
diff changeset
53 :rtype: yielded list
9f2517655a1e Uploaded
cpt
parents:
diff changeset
54 :return: Yields a list of matching features.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
55 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
56 # Either the top level set of [features] or the subfeature attribute
9f2517655a1e Uploaded
cpt
parents:
diff changeset
57 for feature in feature_list:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
58 feature._parent = parent
9f2517655a1e Uploaded
cpt
parents:
diff changeset
59 if not parent:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
60 # Set to self so we cannot go above root.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
61 feature._parent = feature
9f2517655a1e Uploaded
cpt
parents:
diff changeset
62 test_result = test(feature, **test_kwargs)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
63 # if (not invert and test_result) or (invert and not test_result):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
64 if invert ^ test_result:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
65 if not subfeatures:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
66 feature_copy = copy.deepcopy(feature)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
67 feature_copy.sub_features = list()
9f2517655a1e Uploaded
cpt
parents:
diff changeset
68 yield feature_copy
9f2517655a1e Uploaded
cpt
parents:
diff changeset
69 else:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
70 yield feature
9f2517655a1e Uploaded
cpt
parents:
diff changeset
71
9f2517655a1e Uploaded
cpt
parents:
diff changeset
72 if recurse and hasattr(feature, "sub_features"):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
73 for x in feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
74 feature.sub_features,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
75 test,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
76 test_kwargs,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
77 subfeatures=subfeatures,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
78 parent=feature,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
79 invert=invert,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
80 recurse=recurse,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
81 ):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
82 yield x
9f2517655a1e Uploaded
cpt
parents:
diff changeset
83
9f2517655a1e Uploaded
cpt
parents:
diff changeset
84
9f2517655a1e Uploaded
cpt
parents:
diff changeset
85 def fetchParent(feature):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
86 if not hasattr(feature, "_parent") or feature._parent is None:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
87 return feature
9f2517655a1e Uploaded
cpt
parents:
diff changeset
88 else:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
89 return fetchParent(feature._parent)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
90
9f2517655a1e Uploaded
cpt
parents:
diff changeset
91
9f2517655a1e Uploaded
cpt
parents:
diff changeset
92 def feature_test_true(feature, **kwargs):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
93 return True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
94
9f2517655a1e Uploaded
cpt
parents:
diff changeset
95
9f2517655a1e Uploaded
cpt
parents:
diff changeset
96 def feature_test_type(feature, **kwargs):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
97 if "type" in kwargs:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
98 return str(feature.type).upper() == str(kwargs["type"]).upper()
9f2517655a1e Uploaded
cpt
parents:
diff changeset
99 elif "types" in kwargs:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
100 for x in kwargs["types"]:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
101 if str(feature.type).upper() == str(x).upper():
9f2517655a1e Uploaded
cpt
parents:
diff changeset
102 return True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
103 return False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
104 raise Exception("Incorrect feature_test_type call, need type or types")
9f2517655a1e Uploaded
cpt
parents:
diff changeset
105
9f2517655a1e Uploaded
cpt
parents:
diff changeset
106
9f2517655a1e Uploaded
cpt
parents:
diff changeset
107 def feature_test_qual_value(feature, **kwargs):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
108 """Test qualifier values.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
109
9f2517655a1e Uploaded
cpt
parents:
diff changeset
110 For every feature, check that at least one value in
9f2517655a1e Uploaded
cpt
parents:
diff changeset
111 feature.quailfiers(kwargs['qualifier']) is in kwargs['attribute_list']
9f2517655a1e Uploaded
cpt
parents:
diff changeset
112 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
113 if isinstance(kwargs["qualifier"], list):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
114 for qualifier in kwargs["qualifier"]:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
115 for attribute_value in feature.qualifiers.get(qualifier, []):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
116 if attribute_value in kwargs["attribute_list"]:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
117 return True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
118 else:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
119 for attribute_value in feature.qualifiers.get(kwargs["qualifier"], []):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
120 if attribute_value in kwargs["attribute_list"]:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
121 return True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
122 return False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
123
9f2517655a1e Uploaded
cpt
parents:
diff changeset
124
9f2517655a1e Uploaded
cpt
parents:
diff changeset
125 def feature_test_location(feature, **kwargs):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
126 if "strand" in kwargs:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
127 if feature.location.strand != kwargs["strand"]:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
128 return False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
129
9f2517655a1e Uploaded
cpt
parents:
diff changeset
130 return feature.location.start <= kwargs["loc"] <= feature.location.end
9f2517655a1e Uploaded
cpt
parents:
diff changeset
131
9f2517655a1e Uploaded
cpt
parents:
diff changeset
132
9f2517655a1e Uploaded
cpt
parents:
diff changeset
133 def feature_test_quals(feature, **kwargs):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
134 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
135 Example::
9f2517655a1e Uploaded
cpt
parents:
diff changeset
136
9f2517655a1e Uploaded
cpt
parents:
diff changeset
137 a = Feature(qualifiers={'Note': ['Some notes', 'Aasdf']})
9f2517655a1e Uploaded
cpt
parents:
diff changeset
138
9f2517655a1e Uploaded
cpt
parents:
diff changeset
139 # Check if a contains a Note
9f2517655a1e Uploaded
cpt
parents:
diff changeset
140 feature_test_quals(a, {'Note': None}) # Returns True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
141 feature_test_quals(a, {'Product': None}) # Returns False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
142
9f2517655a1e Uploaded
cpt
parents:
diff changeset
143 # Check if a contains a note with specific value
9f2517655a1e Uploaded
cpt
parents:
diff changeset
144 feature_test_quals(a, {'Note': ['ome']}) # Returns True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
145
9f2517655a1e Uploaded
cpt
parents:
diff changeset
146 # Check if a contains a note with specific value
9f2517655a1e Uploaded
cpt
parents:
diff changeset
147 feature_test_quals(a, {'Note': ['other']}) # Returns False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
148 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
149 for key in kwargs:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
150 if key not in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
151 return False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
152
9f2517655a1e Uploaded
cpt
parents:
diff changeset
153 # Key is present, no value specified
9f2517655a1e Uploaded
cpt
parents:
diff changeset
154 if kwargs[key] is None:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
155 return True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
156
9f2517655a1e Uploaded
cpt
parents:
diff changeset
157 # Otherwise there is a key value we're looking for.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
158 # so we make a list of matches
9f2517655a1e Uploaded
cpt
parents:
diff changeset
159 matches = []
9f2517655a1e Uploaded
cpt
parents:
diff changeset
160 # And check all of the feature qualifier valuse
9f2517655a1e Uploaded
cpt
parents:
diff changeset
161 for value in feature.qualifiers[key]:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
162 # For that kwargs[key] value
9f2517655a1e Uploaded
cpt
parents:
diff changeset
163 for x in kwargs[key]:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
164 matches.append(x in value)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
165
9f2517655a1e Uploaded
cpt
parents:
diff changeset
166 # If none matched, then we return false.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
167 if not any(matches):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
168 return False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
169
9f2517655a1e Uploaded
cpt
parents:
diff changeset
170 return True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
171
9f2517655a1e Uploaded
cpt
parents:
diff changeset
172
9f2517655a1e Uploaded
cpt
parents:
diff changeset
173 def feature_test_contains(feature, **kwargs):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
174 if "index" in kwargs:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
175 return feature.location.start < kwargs["index"] < feature.location.end
9f2517655a1e Uploaded
cpt
parents:
diff changeset
176 elif "range" in kwargs:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
177 return (
9f2517655a1e Uploaded
cpt
parents:
diff changeset
178 feature.location.start < kwargs["range"]["start"] < feature.location.end
9f2517655a1e Uploaded
cpt
parents:
diff changeset
179 and feature.location.start < kwargs["range"]["end"] < feature.location.end
9f2517655a1e Uploaded
cpt
parents:
diff changeset
180 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
181 else:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
182 raise RuntimeError("Must use index or range keyword")
9f2517655a1e Uploaded
cpt
parents:
diff changeset
183
9f2517655a1e Uploaded
cpt
parents:
diff changeset
184
9f2517655a1e Uploaded
cpt
parents:
diff changeset
185 def get_id(feature=None, parent_prefix=None):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
186 result = ""
9f2517655a1e Uploaded
cpt
parents:
diff changeset
187 if parent_prefix is not None:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
188 result += parent_prefix + "|"
9f2517655a1e Uploaded
cpt
parents:
diff changeset
189 if "locus_tag" in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
190 result += feature.qualifiers["locus_tag"][0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
191 elif "gene" in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
192 result += feature.qualifiers["gene"][0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
193 elif "Gene" in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
194 result += feature.qualifiers["Gene"][0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
195 elif "product" in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
196 result += feature.qualifiers["product"][0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
197 elif "Product" in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
198 result += feature.qualifiers["Product"][0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
199 elif "Name" in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
200 result += feature.qualifiers["Name"][0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
201 else:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
202 return feature.id
9f2517655a1e Uploaded
cpt
parents:
diff changeset
203 # Leaving in case bad things happen.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
204 # result += '%s_%s_%s_%s' % (
9f2517655a1e Uploaded
cpt
parents:
diff changeset
205 # feature.id,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
206 # feature.location.start,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
207 # feature.location.end,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
208 # feature.location.strand
9f2517655a1e Uploaded
cpt
parents:
diff changeset
209 # )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
210 return result
9f2517655a1e Uploaded
cpt
parents:
diff changeset
211
9f2517655a1e Uploaded
cpt
parents:
diff changeset
212
9f2517655a1e Uploaded
cpt
parents:
diff changeset
213 def get_gff3_id(gene):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
214 return gene.qualifiers.get("Name", [gene.id])[0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
215
9f2517655a1e Uploaded
cpt
parents:
diff changeset
216
9f2517655a1e Uploaded
cpt
parents:
diff changeset
217 def ensure_location_in_bounds(start=0, end=0, parent_length=0):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
218 # This prevents frameshift errors
9f2517655a1e Uploaded
cpt
parents:
diff changeset
219 while start < 0:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
220 start += 3
9f2517655a1e Uploaded
cpt
parents:
diff changeset
221 while end < 0:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
222 end += 3
9f2517655a1e Uploaded
cpt
parents:
diff changeset
223 while start > parent_length:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
224 start -= 3
9f2517655a1e Uploaded
cpt
parents:
diff changeset
225 while end > parent_length:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
226 end -= 3
9f2517655a1e Uploaded
cpt
parents:
diff changeset
227 return (start, end)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
228
9f2517655a1e Uploaded
cpt
parents:
diff changeset
229
9f2517655a1e Uploaded
cpt
parents:
diff changeset
230 def coding_genes(feature_list):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
231 for x in genes(feature_list):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
232 if (
9f2517655a1e Uploaded
cpt
parents:
diff changeset
233 len(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
234 list(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
235 feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
236 x.sub_features,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
237 feature_test_type,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
238 {"type": "CDS"},
9f2517655a1e Uploaded
cpt
parents:
diff changeset
239 subfeatures=False,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
240 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
241 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
242 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
243 > 0
9f2517655a1e Uploaded
cpt
parents:
diff changeset
244 ):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
245 yield x
9f2517655a1e Uploaded
cpt
parents:
diff changeset
246
9f2517655a1e Uploaded
cpt
parents:
diff changeset
247
9f2517655a1e Uploaded
cpt
parents:
diff changeset
248 def genes(feature_list, feature_type="gene", sort=False):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
249 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
250 Simple filter to extract gene features from the feature set.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
251 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
252
9f2517655a1e Uploaded
cpt
parents:
diff changeset
253 if not sort:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
254 for x in feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
255 feature_list, feature_test_type, {"type": feature_type}, subfeatures=True
9f2517655a1e Uploaded
cpt
parents:
diff changeset
256 ):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
257 yield x
9f2517655a1e Uploaded
cpt
parents:
diff changeset
258 else:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
259 data = list(genes(feature_list, feature_type=feature_type, sort=False))
9f2517655a1e Uploaded
cpt
parents:
diff changeset
260 data = sorted(data, key=lambda feature: feature.location.start)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
261 for x in data:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
262 yield x
9f2517655a1e Uploaded
cpt
parents:
diff changeset
263
9f2517655a1e Uploaded
cpt
parents:
diff changeset
264
9f2517655a1e Uploaded
cpt
parents:
diff changeset
265 def wa_unified_product_name(feature):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
266 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
267 Try and figure out a name. We gave conflicting instructions, so
9f2517655a1e Uploaded
cpt
parents:
diff changeset
268 this isn't as trivial as it should be. Sometimes it will be in
9f2517655a1e Uploaded
cpt
parents:
diff changeset
269 'product' or 'Product', othertimes in 'Name'
9f2517655a1e Uploaded
cpt
parents:
diff changeset
270 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
271 # Manually applied tags.
9f2517655a1e Uploaded
cpt
parents:
diff changeset
272 protein_product = feature.qualifiers.get(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
273 "product", feature.qualifiers.get("Product", [None])
9f2517655a1e Uploaded
cpt
parents:
diff changeset
274 )[0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
275
9f2517655a1e Uploaded
cpt
parents:
diff changeset
276 # If neither of those are available ...
9f2517655a1e Uploaded
cpt
parents:
diff changeset
277 if protein_product is None:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
278 # And there's a name...
9f2517655a1e Uploaded
cpt
parents:
diff changeset
279 if "Name" in feature.qualifiers:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
280 if not is_uuid(feature.qualifiers["Name"][0]):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
281 protein_product = feature.qualifiers["Name"][0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
282
9f2517655a1e Uploaded
cpt
parents:
diff changeset
283 return protein_product
9f2517655a1e Uploaded
cpt
parents:
diff changeset
284
9f2517655a1e Uploaded
cpt
parents:
diff changeset
285
9f2517655a1e Uploaded
cpt
parents:
diff changeset
286 def is_uuid(name):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
287 return name.count("-") == 4 and len(name) == 36
9f2517655a1e Uploaded
cpt
parents:
diff changeset
288
9f2517655a1e Uploaded
cpt
parents:
diff changeset
289
9f2517655a1e Uploaded
cpt
parents:
diff changeset
290 def get_rbs_from(gene):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
291 # Normal RBS annotation types
9f2517655a1e Uploaded
cpt
parents:
diff changeset
292 rbs_rbs = list(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
293 feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
294 gene.sub_features, feature_test_type, {"type": "RBS"}, subfeatures=False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
295 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
296 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
297 rbs_sds = list(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
298 feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
299 gene.sub_features,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
300 feature_test_type,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
301 {"type": "Shine_Dalgarno_sequence"},
9f2517655a1e Uploaded
cpt
parents:
diff changeset
302 subfeatures=False,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
303 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
304 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
305 # Fraking apollo
9f2517655a1e Uploaded
cpt
parents:
diff changeset
306 apollo_exons = list(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
307 feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
308 gene.sub_features, feature_test_type, {"type": "exon"}, subfeatures=False
9f2517655a1e Uploaded
cpt
parents:
diff changeset
309 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
310 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
311 apollo_exons = [x for x in apollo_exons if len(x) < 10]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
312 # These are more NCBI's style
9f2517655a1e Uploaded
cpt
parents:
diff changeset
313 regulatory_elements = list(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
314 feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
315 gene.sub_features,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
316 feature_test_type,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
317 {"type": "regulatory"},
9f2517655a1e Uploaded
cpt
parents:
diff changeset
318 subfeatures=False,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
319 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
320 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
321 rbs_regulatory = list(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
322 feature_lambda(
9f2517655a1e Uploaded
cpt
parents:
diff changeset
323 regulatory_elements,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
324 feature_test_quals,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
325 {"regulatory_class": ["ribosome_binding_site"]},
9f2517655a1e Uploaded
cpt
parents:
diff changeset
326 subfeatures=False,
9f2517655a1e Uploaded
cpt
parents:
diff changeset
327 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
328 )
9f2517655a1e Uploaded
cpt
parents:
diff changeset
329 # Here's hoping you find just one ;)
9f2517655a1e Uploaded
cpt
parents:
diff changeset
330 return rbs_rbs + rbs_sds + rbs_regulatory + apollo_exons
9f2517655a1e Uploaded
cpt
parents:
diff changeset
331
9f2517655a1e Uploaded
cpt
parents:
diff changeset
332
9f2517655a1e Uploaded
cpt
parents:
diff changeset
333 def nice_name(record):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
334 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
335 get the real name rather than NCBI IDs and so on. If fails, will return record.id
9f2517655a1e Uploaded
cpt
parents:
diff changeset
336 """
9f2517655a1e Uploaded
cpt
parents:
diff changeset
337 name = record.id
9f2517655a1e Uploaded
cpt
parents:
diff changeset
338 likely_parental_contig = list(genes(record.features, feature_type="contig"))
9f2517655a1e Uploaded
cpt
parents:
diff changeset
339 if len(likely_parental_contig) == 1:
9f2517655a1e Uploaded
cpt
parents:
diff changeset
340 name = likely_parental_contig[0].qualifiers.get("organism", [name])[0]
9f2517655a1e Uploaded
cpt
parents:
diff changeset
341 return name
9f2517655a1e Uploaded
cpt
parents:
diff changeset
342
9f2517655a1e Uploaded
cpt
parents:
diff changeset
343
9f2517655a1e Uploaded
cpt
parents:
diff changeset
344 def fsort(it):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
345 for i in sorted(it, key=lambda x: int(x.location.start)):
9f2517655a1e Uploaded
cpt
parents:
diff changeset
346 yield i