hdallatorre commited on
Commit
b47b74e
1 Parent(s): 25aa1e9

feat: Commit the configurations file

Browse files
nucleotide_transformer_downstream_tasks.py ADDED
@@ -0,0 +1,149 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright 2020 The HuggingFace Datasets Authors and the current dataset script
2
+ # contributor.
3
+ #
4
+ # Licensed under the Apache License, Version 2.0 (the "License");
5
+ # you may not use this file except in compliance with the License.
6
+ # You may obtain a copy of the License at
7
+ #
8
+ # http://www.apache.org/licenses/LICENSE-2.0
9
+ #
10
+ # Unless required by applicable law or agreed to in writing, software
11
+ # distributed under the License is distributed on an "AS IS" BASIS,
12
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
+ # See the License for the specific language governing permissions and
14
+ # limitations under the License.
15
+ """Script for the dataset containing all the downstream tasks from the Nucleotide
16
+ Transformer paper. There are 18 downstream tasks."""
17
+
18
+ from typing import List
19
+ import datasets
20
+ from Bio import SeqIO
21
+
22
+ # Find for instance the citation on arxiv or on the dataset repo/website
23
+ _CITATION = """\
24
+ @article{dalla2023nucleotide,
25
+ title={The Nucleotide Transformer: Building and Evaluating Robust Foundation Models for Human Genomics},
26
+ author={Dalla-Torre, Hugo and Gonzalez, Liam and Mendoza-Revilla, Javier and Carranza, Nicolas Lopez and Grzywaczewski, Adam Henryk and Oteri, Francesco and Dallago, Christian and Trop, Evan and Sirelkhatim, Hassan and Richard, Guillaume and others},
27
+ journal={bioRxiv},
28
+ pages={2023--01},
29
+ year={2023},
30
+ publisher={Cold Spring Harbor Laboratory}
31
+ }
32
+ """
33
+
34
+ # You can copy an official description
35
+ _DESCRIPTION = """\
36
+ 18 classification downstream tasks from the Nucleotide Transformer paper. Each task
37
+ corresponds to a dataset configuration.
38
+ """
39
+
40
+ _HOMEPAGE = "https://github.com/instadeepai/nucleotide-transformer"
41
+
42
+ _LICENSE = "https://github.com/instadeepai/nucleotide-transformer/LICENSE.md"
43
+
44
+ _TASKS = [
45
+ 'H4ac',
46
+ 'H3K36me3',
47
+ 'splice_sites_donors',
48
+ 'splice_sites_acceptors',
49
+ 'H3',
50
+ 'H4',
51
+ 'H3K4me3',
52
+ 'splice_sites_all',
53
+ 'H3K4me1',
54
+ 'H3K14ac',
55
+ 'enhancers_types',
56
+ 'promoter_no_tata',
57
+ 'H3K79me3',
58
+ 'H3K4me2',
59
+ 'promoter_tata',
60
+ 'enhancers',
61
+ 'H3K9ac',
62
+ 'promoter_all'
63
+ ]
64
+
65
+
66
+ class NucleotideTransformerDownstreamTasksConfig(datasets.BuilderConfig):
67
+ """BuilderConfig for The Nucleotide Transformer downstream taks dataset."""
68
+
69
+ def __init__(self, *args, task: str, **kwargs):
70
+ """BuilderConfig downstream tasks dataset.
71
+ Args:
72
+ task (:obj:`str`): Task name.
73
+ **kwargs: keyword arguments forwarded to super.
74
+ """
75
+ super().__init__(
76
+ *args,
77
+ name=f'{task}',
78
+ **kwargs,
79
+ )
80
+ self.task = task
81
+
82
+
83
+ class NucleotideTransformerDownstreamTasks(datasets.GeneratorBasedBuilder):
84
+ """Genomes from 850 species, filtered and split into chunks of consecutive
85
+ nucleotides. 50 genomes are taken for test, 50 for validation and 800
86
+ for training."""
87
+
88
+ VERSION = datasets.Version("1.1.0")
89
+ BUILDER_CONFIG_CLASS = NucleotideTransformerDownstreamTasksConfig
90
+ BUILDER_CONFIGS = [
91
+ NucleotideTransformerDownstreamTasksConfig(task=task) for task in _TASKS
92
+ ]
93
+ DEFAULT_CONFIG_NAME = "enhancers"
94
+
95
+ def _info(self):
96
+
97
+ features = datasets.Features(
98
+ {
99
+ "sequence": datasets.Value("string"),
100
+ "name": datasets.Value("string"),
101
+ "label": datasets.Value("int32"),
102
+ }
103
+ )
104
+ return datasets.DatasetInfo(
105
+ # This is the description that will appear on the datasets page.
106
+ description=_DESCRIPTION,
107
+ # This defines the different columns of the dataset and their types
108
+ features=features,
109
+ # Homepage of the dataset for documentation
110
+ homepage=_HOMEPAGE,
111
+ # License for the dataset if available
112
+ license=_LICENSE,
113
+ # Citation for the dataset
114
+ citation=_CITATION,
115
+ )
116
+
117
+ def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
118
+
119
+ train_file = dl_manager.download_and_extract(self.config.task + '/train.fna')
120
+ test_file = dl_manager.download_and_extract(self.config.task + '/test.fna')
121
+
122
+ return [
123
+ datasets.SplitGenerator(name=datasets.Split.TRAIN,
124
+ gen_kwargs={"file": train_file}
125
+ ),
126
+ datasets.SplitGenerator(name=datasets.Split.TEST,
127
+ gen_kwargs={"file": test_file}
128
+ ),
129
+ ]
130
+
131
+ # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
132
+ def _generate_examples(self, file):
133
+ key = 0
134
+ with open(file, 'rt') as f:
135
+ fasta_sequences = SeqIO.parse(f, 'fasta')
136
+
137
+ for record in fasta_sequences:
138
+
139
+ # parse descriptions in the fasta file
140
+ sequence, name = str(record.seq), str(record.name)
141
+ label = int(name.split("|")[-1])
142
+
143
+ # yield example
144
+ yield key, {
145
+ 'sequence': sequence,
146
+ 'name': name,
147
+ 'label': label,
148
+ }
149
+ key += 1