1
|
|
|
import time |
2
|
|
|
|
3
|
|
|
from pyensemblrest import EnsemblRest |
4
|
|
|
|
5
|
|
|
# Setup a new EnsemblRest object |
6
|
|
|
ensRest = EnsemblRest() |
7
|
|
|
|
8
|
|
|
# Check the user agent |
9
|
|
|
print(ensRest.get_user_agent()) |
10
|
|
|
|
11
|
|
|
# Archive |
12
|
|
|
print(ensRest.getArchiveById(id="ENSG00000157764")) |
13
|
|
|
print(ensRest.getArchiveByMultipleIds(id=["ENSG00000157764", "ENSG00000248378"])) |
14
|
|
|
time.sleep(15) |
15
|
|
|
|
16
|
|
|
# Comparative Genomics |
17
|
|
|
print(ensRest.getCafeGeneTreeById(id="ENSGT00390000003602")) |
18
|
|
|
print(ensRest.getCafeGeneTreeMemberBySymbol(species="human", symbol="BRCA2")) |
19
|
|
|
print(ensRest.getCafeGeneTreeMemberById(species="human", id="ENSG00000167664")) |
20
|
|
|
print(ensRest.getGeneTreeById(id="ENSGT00390000003602")) |
21
|
|
|
print(ensRest.getGeneTreeMemberBySymbol(species="human", symbol="BRCA2")) |
22
|
|
|
print(ensRest.getGeneTreeMemberById(species="human", id="ENSG00000167664")) |
23
|
|
|
print( |
24
|
|
|
ensRest.getAlignmentByRegion( |
25
|
|
|
species="human", |
26
|
|
|
region="X:1000000..1000100:1", |
27
|
|
|
species_set_group="mammals", |
28
|
|
|
) |
29
|
|
|
) |
30
|
|
|
print(ensRest.getHomologyById(species="human", id="ENSG00000157764")) |
31
|
|
|
print(ensRest.getHomologyBySymbol(species="human", symbol="BRCA2")) |
32
|
|
|
time.sleep(15) |
33
|
|
|
|
34
|
|
|
# Cross References |
35
|
|
|
print(ensRest.getXrefsBySymbol(species="human", symbol="BRCA2")) |
36
|
|
|
print(ensRest.getXrefsById(id="ENSG00000157764")) |
37
|
|
|
print(ensRest.getXrefsByName(species="human", name="BRCA2")) |
38
|
|
|
time.sleep(15) |
39
|
|
|
|
40
|
|
|
# Information |
41
|
|
|
print(ensRest.getInfoAnalysis(species="homo_sapiens")) |
42
|
|
|
print( |
43
|
|
|
ensRest.getInfoAssembly(species="homo_sapiens", bands=1) |
44
|
|
|
) # bands is an optional parameter |
45
|
|
|
print(ensRest.getInfoAssemblyRegion(species="homo_sapiens", region_name="X")) |
46
|
|
|
ensRest.timeout = 300 |
47
|
|
|
print(ensRest.getInfoBiotypes(species="homo_sapiens")) # this keeps timing out |
48
|
|
|
ensRest.timeout = 60 |
49
|
|
|
print(ensRest.getInfoBiotypesByGroup(group="coding", object_type="gene")) |
50
|
|
|
print(ensRest.getInfoBiotypesByName(name="protein_coding", object_type="gene")) |
51
|
|
|
print(ensRest.getInfoComparaMethods()) |
52
|
|
|
print(ensRest.getInfoComparaSpeciesSets(methods="EPO")) |
53
|
|
|
print(ensRest.getInfoComparas()) |
54
|
|
|
print(ensRest.getInfoData()) |
55
|
|
|
print(ensRest.getInfoEgVersion()) |
56
|
|
|
print(ensRest.getInfoExternalDbs(species="homo_sapiens")) |
57
|
|
|
print(ensRest.getInfoDivisions()) |
58
|
|
|
print(ensRest.getInfoGenomesByName(name="arabidopsis_thaliana")) |
59
|
|
|
print(ensRest.getInfoGenomesByAccession(accession="U00096")) |
60
|
|
|
print(ensRest.getInfoGenomesByAssembly(assembly_id="GCA_902167145.1")) |
61
|
|
|
print(ensRest.getInfoGenomesByDivision(division="EnsemblPlants")) |
62
|
|
|
print(ensRest.getInfoGenomesByTaxonomy(taxon_name="Homo sapiens")) |
63
|
|
|
print(ensRest.getInfoPing()) |
64
|
|
|
print(ensRest.getInfoRest()) |
65
|
|
|
print(ensRest.getInfoSoftware()) |
66
|
|
|
print(ensRest.getInfoSpecies()) |
67
|
|
|
print(ensRest.getInfoVariationBySpecies(species="homo_sapiens")) |
68
|
|
|
print(ensRest.getInfoVariationConsequenceTypes()) |
69
|
|
|
print( |
70
|
|
|
ensRest.getInfoVariationPopulationIndividuals( |
71
|
|
|
species="human", population_name="1000GENOMES:phase_3:ASW" |
72
|
|
|
) |
73
|
|
|
) |
74
|
|
|
print(ensRest.getInfoVariationPopulations(species="homo_sapiens", filter="LD")) |
75
|
|
|
time.sleep(15) |
76
|
|
|
|
77
|
|
|
# Linkage Disequilibrium |
78
|
|
|
print( |
79
|
|
|
ensRest.getLdId( |
80
|
|
|
species="homo_sapiens", |
81
|
|
|
id="rs56116432", |
82
|
|
|
population_name="1000GENOMES:phase_3:KHV", |
83
|
|
|
window_size=500, |
84
|
|
|
d_prime=1.0, |
85
|
|
|
) |
86
|
|
|
) |
87
|
|
|
print(ensRest.getLdPairwise(species="homo_sapiens", id1="rs6792369", id2="rs1042779")) |
88
|
|
|
print( |
89
|
|
|
ensRest.getLdRegion( |
90
|
|
|
species="homo_sapiens", |
91
|
|
|
region="6:25837556..25843455", |
92
|
|
|
population_name="1000GENOMES:phase_3:KHV", |
93
|
|
|
) |
94
|
|
|
) |
95
|
|
|
time.sleep(15) |
96
|
|
|
|
97
|
|
|
# Lookup |
98
|
|
|
print(ensRest.getLookupById(id="ENSG00000157764")) |
99
|
|
|
print(ensRest.getLookupByMultipleIds(ids=["ENSG00000157764", "ENSG00000248378"])) |
100
|
|
|
print(ensRest.getLookupBySymbol(species="homo_sapiens", symbol="BRCA2", expand=1)) |
101
|
|
|
print( |
102
|
|
|
ensRest.getLookupByMultipleSymbols( |
103
|
|
|
species="homo_sapiens", symbols=["BRCA2", "BRAF"] |
104
|
|
|
) |
105
|
|
|
) |
106
|
|
|
time.sleep(15) |
107
|
|
|
|
108
|
|
|
# Mapping |
109
|
|
|
print(ensRest.getMapCdnaToRegion(id="ENST00000288602", region="100..300")) |
110
|
|
|
print(ensRest.getMapCdsToRegion(id="ENST00000288602", region="1..1000")) |
111
|
|
|
print( |
112
|
|
|
ensRest.getMapAssemblyOneToTwo( |
113
|
|
|
species="homo_sapiens", |
114
|
|
|
asm_one="GRCh37", |
115
|
|
|
region="X:1000000..1000100:1", |
116
|
|
|
asm_two="GRCh38", |
117
|
|
|
) |
118
|
|
|
) |
119
|
|
|
print(ensRest.getMapTranslationToRegion(id="ENSP00000288602", region="100..300")) |
120
|
|
|
time.sleep(15) |
121
|
|
|
|
122
|
|
|
# Ontologies and Taxonomy |
123
|
|
|
print(ensRest.getAncestorsById(id="GO:0005667")) |
124
|
|
|
print(ensRest.getAncestorsChartById(id="GO:0005667")) |
125
|
|
|
print(ensRest.getDescendantsById(id="GO:0005667")) |
126
|
|
|
print(ensRest.getOntologyById(id="GO:0005667")) |
127
|
|
|
print(ensRest.getOntologyByName(name="transcription factor complex")) |
128
|
|
|
print(ensRest.getTaxonomyClassificationById(id="9606")) |
129
|
|
|
print(ensRest.getTaxonomyById(id="9606")) |
130
|
|
|
print(ensRest.getTaxonomyByName(name="Homo%25")) |
131
|
|
|
time.sleep(15) |
132
|
|
|
|
133
|
|
|
# Overlap |
134
|
|
|
print(ensRest.getOverlapById(id="ENSG00000157764", feature="gene")) |
135
|
|
|
print( |
136
|
|
|
ensRest.getOverlapByRegion( |
137
|
|
|
species="homo_sapiens", region="X:1..1000:1", feature="gene" |
138
|
|
|
) |
139
|
|
|
) |
140
|
|
|
print(ensRest.getOverlapByTranslation(id="ENSP00000288602")) |
141
|
|
|
time.sleep(15) |
142
|
|
|
|
143
|
|
|
# Phenotype annotations |
144
|
|
|
print(ensRest.getPhenotypeByAccession(species="homo_sapiens", accession="EFO:0003900")) |
145
|
|
|
print(ensRest.getPhenotypeByGene(species="homo_sapiens", gene="ENSG00000157764")) |
146
|
|
|
print( |
147
|
|
|
ensRest.getPhenotypeByRegion(species="homo_sapiens", region="9:22125500-22136000:1") |
148
|
|
|
) |
149
|
|
|
print(ensRest.getPhenotypeByTerm(species="homo_sapiens", term="coffee consumption")) |
150
|
|
|
time.sleep(15) |
151
|
|
|
|
152
|
|
|
# Regulation |
153
|
|
|
print( |
154
|
|
|
ensRest.getRegulationBindingMatrix( |
155
|
|
|
species="homo_sapiens", binding_matrix="ENSPFM0001" |
156
|
|
|
) |
157
|
|
|
) |
158
|
|
|
time.sleep(15) |
159
|
|
|
|
160
|
|
|
# Sequences |
161
|
|
|
print(ensRest.getSequenceById(id="ENSG00000157764")) |
162
|
|
|
print(ensRest.getSequenceByMultipleIds(ids=["ENSG00000157764", "ENSG00000248378"])) |
163
|
|
|
print( |
164
|
|
|
ensRest.getSequenceByRegion(species="homo_sapiens", region="X:1000000..1000100:1") |
165
|
|
|
) |
166
|
|
|
print( |
167
|
|
|
ensRest.getSequenceByMultipleRegions( |
168
|
|
|
species="homo_sapiens", |
169
|
|
|
regions=["X:1000000..1000100:1", "ABBA01004489.1:1..100"], |
170
|
|
|
) |
171
|
|
|
) |
172
|
|
|
time.sleep(15) |
173
|
|
|
|
174
|
|
|
# Transcript Haplotypes |
175
|
|
|
print(ensRest.getTranscriptHaplotypes(species="homo_sapiens", id="ENST00000288602")) |
176
|
|
|
time.sleep(15) |
177
|
|
|
|
178
|
|
|
# VEP |
179
|
|
|
print( |
180
|
|
|
ensRest.getVariantConsequencesByHGVSNotation( |
181
|
|
|
species="homo_sapiens", hgvs_notation="ENST00000366667:c.803C>T" |
182
|
|
|
) |
183
|
|
|
) |
184
|
|
|
print( |
185
|
|
|
ensRest.getVariantConsequencesByMultipleHGVSNotations( |
186
|
|
|
species="homo_sapiens", |
187
|
|
|
hgvs_notations=["ENST00000366667:c.803C>T", "9:g.22125504G>C"], |
188
|
|
|
) |
189
|
|
|
) |
190
|
|
|
print(ensRest.getVariantConsequencesById(species="homo_sapiens", id="rs56116432")) |
191
|
|
|
print( |
192
|
|
|
ensRest.getVariantConsequencesByMultipleIds( |
193
|
|
|
species="homo_sapiens", ids=["rs56116432", "COSM476", "__VAR(sv_id)__"] |
194
|
|
|
) |
195
|
|
|
) |
196
|
|
|
print( |
197
|
|
|
ensRest.getVariantConsequencesByRegion( |
198
|
|
|
species="homo_sapiens", region="9:22125503-22125502:1", allele="C" |
199
|
|
|
) |
200
|
|
|
) |
201
|
|
|
print( |
202
|
|
|
ensRest.getVariantConsequencesByMultipleRegions( |
203
|
|
|
species="homo_sapiens", |
204
|
|
|
variants=[ |
205
|
|
|
"21 26960070 rs116645811 G A . . .", |
206
|
|
|
"21 26965148 rs1135638 G A . . .", |
207
|
|
|
], |
208
|
|
|
) |
209
|
|
|
) |
210
|
|
|
time.sleep(15) |
211
|
|
|
|
212
|
|
|
# Variation |
213
|
|
|
print(ensRest.getVariationRecoderById(species="homo_sapiens", id="rs56116432")) |
214
|
|
|
print( |
215
|
|
|
ensRest.getVariationRecoderByMultipleIds( |
216
|
|
|
species="homo_sapiens", ids=["rs56116432", "rs1042779"] |
217
|
|
|
) |
218
|
|
|
) |
219
|
|
|
print(ensRest.getVariationById(species="homo_sapiens", id="rs56116432")) |
220
|
|
|
print(ensRest.getVariationByPMCID(species="homo_sapiens", pmcid="PMC5002951")) |
221
|
|
|
print(ensRest.getVariationByPMID(species="homo_sapiens", pmid="26318936")) |
222
|
|
|
print( |
223
|
|
|
ensRest.getVariationByMultipleIds( |
224
|
|
|
species="homo_sapiens", ids=["rs56116432", "COSM476", "__VAR(sv_id)__"] |
225
|
|
|
) |
226
|
|
|
) |
227
|
|
|
time.sleep(15) |
228
|
|
|
|
229
|
|
|
# Variation GA4GH |
230
|
|
|
print(ensRest.getGA4GHBeacon()) |
231
|
|
|
print( |
232
|
|
|
ensRest.getGA4GHBeaconQuery( |
233
|
|
|
alternateBases="C", |
234
|
|
|
assemblyId="GRCh38", |
235
|
|
|
end="23125503", |
236
|
|
|
referenceBases="G", |
237
|
|
|
referenceName="9", |
238
|
|
|
start="22125503", |
239
|
|
|
variantType="DUP", |
240
|
|
|
) |
241
|
|
|
) |
242
|
|
|
print( |
243
|
|
|
ensRest.postGA4GHBeaconQuery( |
244
|
|
|
alternateBases="C", |
245
|
|
|
assemblyId="GRCh38", |
246
|
|
|
end="23125503", |
247
|
|
|
referenceBases="G", |
248
|
|
|
referenceName="9", |
249
|
|
|
start="22125503", |
250
|
|
|
variantType="DUP", |
251
|
|
|
) |
252
|
|
|
) |
253
|
|
|
print(ensRest.getGA4GHFeaturesById(id="ENST00000408937.7")) |
254
|
|
|
ensRest.timeout = 180 |
255
|
|
|
print( |
256
|
|
|
ensRest.searchGA4GHFeatures( |
257
|
|
|
parentId="ENST00000408937.7", |
258
|
|
|
featureSetId="", |
259
|
|
|
featureTypes=["cds"], |
260
|
|
|
end=220023, |
261
|
|
|
referenceName="X", |
262
|
|
|
start=197859, |
263
|
|
|
pageSize=1, |
264
|
|
|
) |
265
|
|
|
) # this keeps timing out |
266
|
|
|
ensRest.timeout = 60 |
267
|
|
|
print(ensRest.searchGA4GHCallset(variantSetId=1, pageSize=2)) |
268
|
|
|
print(ensRest.getGA4GHCallsetById(id="1")) |
269
|
|
|
print(ensRest.searchGA4GHDatasets(pageSize=3)) |
270
|
|
|
print(ensRest.getGA4GHDatasetsById(id="6e340c4d1e333c7a676b1710d2e3953c")) |
271
|
|
|
print(ensRest.searchGA4GHFeaturesets(datasetId="Ensembl")) |
272
|
|
|
print(ensRest.getGA4GHFeaturesetsById(id="Ensembl")) |
273
|
|
|
print(ensRest.getGA4GHVariantsById(id="1:rs1333049")) |
274
|
|
|
print( |
275
|
|
|
ensRest.searchGA4GHVariantAnnotations( |
276
|
|
|
variantAnnotationSetId="Ensembl", |
277
|
|
|
referenceId="9489ae7581e14efcad134f02afafe26c", |
278
|
|
|
start=25221400, |
279
|
|
|
end=25221500, |
280
|
|
|
pageSize=1, |
281
|
|
|
) |
282
|
|
|
) |
283
|
|
|
print( |
284
|
|
|
ensRest.searchGA4GHVariants( |
285
|
|
|
variantSetId=1, |
286
|
|
|
referenceName=22, |
287
|
|
|
start=25455086, |
288
|
|
|
end=25455087, |
289
|
|
|
pageToken="", |
290
|
|
|
pageSize=1, |
291
|
|
|
) |
292
|
|
|
) |
293
|
|
|
print( |
294
|
|
|
ensRest.searchGA4GHVariantsets( |
295
|
|
|
datasetId="6e340c4d1e333c7a676b1710d2e3953c", pageToken="", pageSize=2 |
296
|
|
|
) |
297
|
|
|
) |
298
|
|
|
print(ensRest.getGA4GHVariantsetsById(id=1)) |
299
|
|
|
print(ensRest.searchGA4GHReferences(referenceSetId="GRCh38", pageSize=10)) |
300
|
|
|
print(ensRest.getGA4GHReferencesById(id="9489ae7581e14efcad134f02afafe26c")) |
301
|
|
|
print(ensRest.searchGA4GHReferencesets()) |
302
|
|
|
print(ensRest.getGA4GHReferencesetsById(id="GRCh38")) |
303
|
|
|
print(ensRest.searchGA4GHVariantAnnotationsets(variantSetId="Ensembl")) |
304
|
|
|
print(ensRest.getGA4GHVariantAnnotationsetsById(id="Ensembl")) |
305
|
|
|
time.sleep(15) |
306
|
|
|
|