monsoon-nlp commited on
Commit
61fa687
·
verified ·
1 Parent(s): 23b3742

Delete plant-multi-species-genomes.py

Browse files
Files changed (1) hide show
  1. plant-multi-species-genomes.py +0 -173
plant-multi-species-genomes.py DELETED
@@ -1,173 +0,0 @@
1
-
2
- """Script for the plant multi-species genomes dataset. This dataset contains the genomes
3
- from 48 different species."""
4
-
5
- from typing import List
6
- import datasets
7
- from Bio import SeqIO
8
- import os
9
-
10
-
11
- # Find for instance the citation on arxiv or on the dataset repo/website
12
- _CITATION = """\
13
- @article{o2016reference,
14
- title={Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation},
15
- author={O'Leary, Nuala A and Wright, Mathew W and Brister, J Rodney and Ciufo, Stacy and Haddad, Diana and McVeigh, Rich and Rajput, Bhanu and Robbertse, Barbara and Smith-White, Brian and Ako-Adjei, Danso and others},
16
- journal={Nucleic acids research},
17
- volume={44},
18
- number={D1},
19
- pages={D733--D745},
20
- year={2016},
21
- publisher={Oxford University Press}
22
- }
23
- """
24
-
25
-
26
- # You can copy an official description
27
- _DESCRIPTION = """\
28
- Dataset made of diverse genomes available on NCBI and coming from 48 different species.
29
- Test and validation are made of 2 species each. The rest of the genomes are used for training.
30
- Default configuration "6kbp" yields chunks of 6.2kbp (100bp overlap on each side). The chunks of DNA are cleaned and processed so that
31
- they can only contain the letters A, T, C, G and N.
32
- """
33
-
34
- _HOMEPAGE = "https://www.ncbi.nlm.nih.gov/"
35
-
36
- _LICENSE = "https://www.ncbi.nlm.nih.gov/home/about/policies/"
37
-
38
- _CHUNK_LENGTHS = [6000,]
39
-
40
-
41
- def filter_fn(char: str) -> str:
42
- """
43
- Transforms any letter different from a base nucleotide into an 'N'.
44
- """
45
- if char in {'A', 'T', 'C', 'G'}:
46
- return char
47
- else:
48
- return 'N'
49
-
50
-
51
- def clean_sequence(seq: str) -> str:
52
- """
53
- Process a chunk of DNA to have all letters in upper and restricted to
54
- A, T, C, G and N.
55
- """
56
- seq = seq.upper()
57
- seq = map(filter_fn, seq)
58
- seq = ''.join(list(seq))
59
- return seq
60
-
61
-
62
- class PlantMultiSpeciesGenomesConfig(datasets.BuilderConfig):
63
- """BuilderConfig for the Plant Multi Species Pre-training Dataset."""
64
-
65
- def __init__(self, *args, chunk_length: int, overlap: int = 100, **kwargs):
66
- """BuilderConfig for the multi species genomes.
67
- Args:
68
- chunk_length (:obj:`int`): Chunk length.
69
- overlap: (:obj:`int`): Overlap in base pairs for two consecutive chunks (defaults to 100).
70
- **kwargs: keyword arguments forwarded to super.
71
- """
72
- num_kbp = int(chunk_length/1000)
73
- super().__init__(
74
- *args,
75
- name=f'{num_kbp}kbp',
76
- **kwargs,
77
- )
78
- self.chunk_length = chunk_length
79
- self.overlap = overlap
80
-
81
-
82
- class PlantMultiSpeciesGenomes(datasets.GeneratorBasedBuilder):
83
- """Genomes from 48 species, filtered and split into chunks of consecutive
84
- nucleotides. 2 genomes are taken for test, 2 for validation and 44
85
- for training."""
86
-
87
- VERSION = datasets.Version("1.1.0")
88
- BUILDER_CONFIG_CLASS = PlantMultiSpeciesGenomesConfig
89
- BUILDER_CONFIGS = [PlantMultiSpeciesGenomesConfig(chunk_length=chunk_length) for chunk_length in _CHUNK_LENGTHS]
90
- DEFAULT_CONFIG_NAME = "6kbp"
91
-
92
- def _info(self):
93
-
94
- features = datasets.Features(
95
- {
96
- "sequence": datasets.Value("string"),
97
- "description": datasets.Value("string"),
98
- "start_pos": datasets.Value("int32"),
99
- "end_pos": datasets.Value("int32"),
100
- }
101
- )
102
- return datasets.DatasetInfo(
103
- # This is the description that will appear on the datasets page.
104
- description=_DESCRIPTION,
105
- # This defines the different columns of the dataset and their types
106
- features=features,
107
- # Homepage of the dataset for documentation
108
- homepage=_HOMEPAGE,
109
- # License for the dataset if available
110
- license=_LICENSE,
111
- # Citation for the dataset
112
- citation=_CITATION,
113
- )
114
-
115
- def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
116
-
117
- filepaths_txt = dl_manager.download_and_extract('plant_genome_file_names.txt')
118
- with open(filepaths_txt) as f:
119
- filepaths = [os.path.join("plant_genomes",filepath.rstrip()) for filepath in f]
120
-
121
- test_paths = filepaths[-2:] # 2 genomes for test set
122
- validation_paths = filepaths[-4:-2] # 2 genomes for validation set
123
- train_paths = filepaths[:-4] # 44 genomes for training
124
-
125
- train_downloaded_files = dl_manager.download_and_extract(train_paths)
126
- test_downloaded_files = dl_manager.download_and_extract(test_paths)
127
- validation_downloaded_files = dl_manager.download_and_extract(validation_paths)
128
-
129
- return [
130
- datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"files": train_downloaded_files, "chunk_length": self.config.chunk_length}),
131
- datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"files": validation_downloaded_files, "chunk_length": self.config.chunk_length}),
132
- datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"files": test_downloaded_files, "chunk_length": self.config.chunk_length}),
133
- ]
134
-
135
- # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
136
- def _generate_examples(self, files, chunk_length):
137
- key = 0
138
- for file in files:
139
- with open(file, 'rt') as f:
140
- fasta_sequences = SeqIO.parse(f, 'fasta')
141
-
142
- for record in fasta_sequences:
143
-
144
- # parse descriptions in the fasta file
145
- sequence, description = str(record.seq), record.description
146
-
147
- # clean chromosome sequence
148
- sequence = clean_sequence(sequence)
149
- seq_length = len(sequence)
150
-
151
- # split into chunks
152
- num_chunks = (seq_length - 2 * self.config.overlap) // chunk_length
153
-
154
- if num_chunks < 1:
155
- continue
156
-
157
- sequence = sequence[:(chunk_length * num_chunks + 2 * self.config.overlap)]
158
- seq_length = len(sequence)
159
-
160
- for i in range(num_chunks):
161
- # get chunk
162
- start_pos = i * chunk_length
163
- end_pos = min(seq_length, (i+1) * chunk_length + 2 * self.config.overlap)
164
- chunk_sequence = sequence[start_pos:end_pos]
165
-
166
- # yield chunk
167
- yield key, {
168
- 'sequence': chunk_sequence,
169
- 'description': description,
170
- 'start_pos': start_pos,
171
- 'end_pos': end_pos,
172
- }
173
- key += 1