diff --git a/backend/metagenedb/api/catalog/views/gene.py b/backend/metagenedb/api/catalog/views/gene.py
index ba1c3ff4511869cc62eb9572b34796faba2886ec..4b23b38f18605a9823a51b6dd27349288f3322bd 100644
--- a/backend/metagenedb/api/catalog/views/gene.py
+++ b/backend/metagenedb/api/catalog/views/gene.py
@@ -52,53 +52,17 @@ class GeneViewSet(BulkViewSet):
response['Content-Disposition'] = 'attachment; filename=%s' % filename
return response
- def _extract_taxonomy_info(self, gene):
- if gene.taxonomy is None:
- return ['', '', '', '']
- return [
- gene.taxonomy.tax_id, gene.taxonomy.name,
- gene.taxonomy.rank, gene.taxonomy.one_line_detailed_taxonomy
- ]
-
- def _extract_function_info(self, gene):
- if not gene.functions.all():
- return ['', '']
- function_ids = {
- 'kegg': [],
- 'eggnog': []
- }
- for function in gene.functions.all():
- function_ids.get(function.source).append(function.function_id)
- return [
- ';'.join(function_ids['kegg']),
- ';'.join(function_ids['eggnog'])
- ]
-
- def _get_metadata_line(self, gene):
- """
- Transform gene content to a line for metadata extract
- """
- gene_items = [
- gene.gene_id, gene.name, gene.source, gene.length,
- ]
- gene_items = gene_items + self._extract_taxonomy_info(gene)
- gene_items = gene_items + self._extract_function_info(gene)
- return ','.join([str(item) for item in gene_items])
-
def _build_csv_response(self):
queryset = self.filter_queryset(self.get_queryset())
queryset = queryset.select_related("taxonomy").prefetch_related("functions")
- # if self._check_too_many_genes(queryset):
- # return self.too_many_genes_error_response
+ if self._check_too_many_genes(queryset):
+ return self.too_many_genes_error_response
with StringIO() as csv_file:
# Write header
- header = ",".join([
- 'gene_id', 'gene_name', 'gene_source', 'length', 'tax_id', 'tax_name', 'tax_rank', 'tax_full'
- 'kegg_id', 'eggnog_id',
- ])
+ header = Gene.CSV_HEADER
csv_file.write(f"{header}\n")
- for gene in queryset:
- csv_file.write(f"{self._get_metadata_line(gene)}\n")
+ for gene in queryset.iterator():
+ csv_file.write(f"{gene.csv}\n")
# generate the file
response = HttpResponse(csv_file.getvalue(), content_type='text/csv')
filename = 'metagenedb.csv'
diff --git a/backend/metagenedb/api/catalog/views/test_gene.py b/backend/metagenedb/api/catalog/views/test_gene.py
index 41977fd91d36a23b392db3a59a0011250e9a17b3..2700261c250996019c44d77399591f0c92978b9d 100644
--- a/backend/metagenedb/api/catalog/views/test_gene.py
+++ b/backend/metagenedb/api/catalog/views/test_gene.py
@@ -3,9 +3,6 @@ from django.urls import reverse
from rest_framework import status
from metagenedb.api.catalog.views import GeneViewSet
-from metagenedb.apps.catalog.factory import (
- GeneFactory, GeneWithEggNOGFactory, GeneWithKeggFactory, GeneWithTaxonomyFactory
-)
class GeneViewSetMock(GeneViewSet):
@@ -26,54 +23,3 @@ class TestGenes(TestCase):
url = reverse('api:catalog:v1:genes-list')
resp = self.client.get(url)
self.assertEqual(resp.status_code, status.HTTP_200_OK)
-
- def test_get_metadata_line_no_functions(self):
- gene = GeneFactory()
- expected_items = [
- gene.gene_id, gene.name, gene.source, gene.length,
- '', '', '', '', '', ''
- ]
- expected_line = ','.join([str(item) for item in expected_items])
- # Make test with method from GeneViewSet
- viewset = GeneViewSetMock()
- tested_line = viewset._get_metadata_line(gene)
- self.assertEqual(tested_line, expected_line)
-
- def test_get_metadata_line_with_taxonomy(self):
- gene = GeneWithTaxonomyFactory()
- expected_items = [
- gene.gene_id, gene.name, gene.source, gene.length,
- gene.taxonomy.tax_id, gene.taxonomy.name, gene.taxonomy.rank, gene.taxonomy.one_line_detailed_taxonomy,
- '', ''
- ]
- expected_line = ','.join([str(item) for item in expected_items])
- # Make test with method from GeneViewSet
- viewset = GeneViewSetMock()
- tested_line = viewset._get_metadata_line(gene)
- self.assertEqual(tested_line, expected_line)
-
- def test_get_metadata_line_with_kegg(self):
- gene = GeneWithKeggFactory()
- expected_items = [
- gene.gene_id, gene.name, gene.source, gene.length,
- '', '', '', '',
- gene.functions.all()[0].function_id, ''
- ]
- expected_line = ','.join([str(item) for item in expected_items])
- # Make test with method from GeneViewSet
- viewset = GeneViewSetMock()
- tested_line = viewset._get_metadata_line(gene)
- self.assertEqual(tested_line, expected_line)
-
- def test_get_metadata_line_with_eggnog(self):
- gene = GeneWithEggNOGFactory()
- expected_items = [
- gene.gene_id, gene.name, gene.source, gene.length,
- '', '', '', '',
- '', gene.functions.all()[0].function_id,
- ]
- expected_line = ','.join([str(item) for item in expected_items])
- # Make test with method from GeneViewSet
- viewset = GeneViewSetMock()
- tested_line = viewset._get_metadata_line(gene)
- self.assertEqual(tested_line, expected_line)
diff --git a/backend/metagenedb/apps/catalog/models/gene.py b/backend/metagenedb/apps/catalog/models/gene.py
index 27bde00f404a0b3523896766bb3ccd835df56d9f..1291e951de442b693921398336c8a82f2bdfbc09 100644
--- a/backend/metagenedb/apps/catalog/models/gene.py
+++ b/backend/metagenedb/apps/catalog/models/gene.py
@@ -1,3 +1,5 @@
+from itertools import repeat
+
from django.db import models
from .function import Function
@@ -12,6 +14,11 @@ class Gene(models.Model):
(IGC, 'IGC'),
(VIRGO, 'Virgo'),
]
+ CSV_HEADER = ','.join([
+ 'gene_id', 'gene_name', 'gene_source', 'length',
+ 'tax_id', 'tax_name', 'tax_rank', 'tax_full',
+ 'kegg_id', 'eggnog_id',
+ ])
gene_id = models.SlugField(max_length=100, db_index=True, unique=True)
name = models.CharField(max_length=100, unique=True)
@@ -25,13 +32,54 @@ class Gene(models.Model):
)
source = models.CharField(max_length=10, choices=SOURCE_CHOICES, default=UNDEFINED)
- def __str__(self):
+ def __str__(self) -> str:
return self.gene_id
@property
- def fasta(self):
+ def fasta(self) -> str:
return f">{self.gene_id}\n{self.sequence}\n"
+ @property
+ def csv_header(self) -> str:
+ return self.CSV_HEADER
+
+ @property
+ def csv_gene(self) -> str:
+ return ",".join([
+ self.name, self.source, str(self.length),
+ ])
+
+ @property
+ def csv_tax(self) -> str:
+ if self.taxonomy is None:
+ return ",".join(list(repeat('', 4)))
+ else:
+ return self.taxonomy.csv
+
+ @property
+ def csv_functions(self) -> str:
+ if not self.functions.all():
+ function_list = list(repeat('', 2))
+ else:
+
+ function_ids = {
+ 'kegg': [],
+ 'eggnog': []
+ }
+ for function in self.functions.all():
+ function_ids.get(function.source).append(function.function_id)
+ function_list = [
+ ';'.join(function_ids['kegg']),
+ ';'.join(function_ids['eggnog'])
+ ]
+ return ",".join(function_list)
+
+ @property
+ def csv(self) -> str:
+ return ",".join(
+ [self.gene_id, self.csv_gene, self.csv_tax, self.csv_functions]
+ )
+
class Meta:
ordering = ['-gene_id']
diff --git a/backend/metagenedb/apps/catalog/models/taxonomy.py b/backend/metagenedb/apps/catalog/models/taxonomy.py
index 459be18f2a3f7a951576308c0fe48dc931fbda2a..058182ecccaa0bed5e972f8f5720ea73026539af 100644
--- a/backend/metagenedb/apps/catalog/models/taxonomy.py
+++ b/backend/metagenedb/apps/catalog/models/taxonomy.py
@@ -42,6 +42,9 @@ class Taxonomy(models.Model):
('varietas', 'Varietas'),
('species_group', 'Species group'),
]
+ CSV_HEADER = ','.join([
+ 'tax_id', 'tax_name', 'tax_rank', 'tax_full',
+ ])
tax_id = models.CharField(max_length=20, unique=True, db_index=True)
name = models.CharField(max_length=200, default=NAME_DEFAULT)
@@ -98,6 +101,14 @@ class Taxonomy(models.Model):
self._one_line_detailed_taxonomy = self._compute_one_line_detailed_taxonomy()
return self._one_line_detailed_taxonomy
+ @property
+ def csv_header(self) -> str:
+ return self.CSV_HEADER
+
+ @property
+ def csv(self) -> str:
+ return ','.join([self.tax_id, self.name, self.rank, self.one_line_detailed_taxonomy])
+
class Meta:
verbose_name_plural = "Taxonomy"
ordering = ['-tax_id']
diff --git a/backend/metagenedb/apps/catalog/models/test_gene.py b/backend/metagenedb/apps/catalog/models/test_gene.py
new file mode 100644
index 0000000000000000000000000000000000000000..d8d173ae30d60424c1759aa3cd624cc6309d2374
--- /dev/null
+++ b/backend/metagenedb/apps/catalog/models/test_gene.py
@@ -0,0 +1,56 @@
+from rest_framework.test import APITestCase
+
+from metagenedb.apps.catalog.factory import (
+ GeneWithEggNOGFactory, GeneWithKeggFactory, GeneWithTaxonomyFactory
+)
+
+
+class TestGeneCSV(APITestCase):
+
+ @classmethod
+ def setUpTestData(cls):
+ """
+ Build some test data for different tests
+ """
+ cls.gene_eggnog = GeneWithEggNOGFactory.create()
+ cls.gene_kegg = GeneWithKeggFactory()
+ cls.gene_tax = GeneWithTaxonomyFactory()
+
+ def test_csv_header(self):
+ expected_header = 'gene_id,gene_name,gene_source,length,tax_id,tax_name,tax_rank,tax_full,kegg_id,eggnog_id'
+ self.assertEqual(self.gene_tax.csv_header, expected_header)
+
+ def test_csv_gene(self):
+ expected = f"{self.gene_tax.name},{self.gene_tax.source},{self.gene_tax.length}"
+ self.assertEqual(self.gene_tax.csv_gene, expected)
+
+ def test_csv_tax(self):
+ expected = (
+ f"{self.gene_tax.taxonomy.tax_id},{self.gene_tax.taxonomy.name},"
+ f"{self.gene_tax.taxonomy.rank},{self.gene_tax.taxonomy.one_line_detailed_taxonomy}"
+ )
+ self.assertEqual(self.gene_tax.csv_tax, expected)
+
+ def test_csv_tax_empty(self):
+ expected = ",,,"
+ self.assertEqual(self.gene_kegg.csv_tax, expected)
+ self.assertEqual(self.gene_eggnog.csv_tax, expected)
+
+ def test_csv_functions(self):
+ expected = f"{self.gene_kegg.functions.all()[0].function_id},"
+ self.assertEqual(self.gene_kegg.csv_functions, expected)
+ expected = f",{self.gene_eggnog.functions.all()[0].function_id}"
+ self.assertEqual(self.gene_eggnog.csv_functions, expected)
+
+ def test_csv_functions_empty(self):
+ expected = ","
+ self.assertEqual(self.gene_tax.csv_functions, expected)
+
+ def test_csv(self):
+ expected = (
+ f"{self.gene_tax.gene_id},{self.gene_tax.name},{self.gene_tax.source},{self.gene_tax.length},"
+ f"{self.gene_tax.taxonomy.tax_id},{self.gene_tax.taxonomy.name},"
+ f"{self.gene_tax.taxonomy.rank},{self.gene_tax.taxonomy.one_line_detailed_taxonomy},"
+ f","
+ )
+ self.assertEqual(self.gene_tax.csv, expected)
diff --git a/backend/metagenedb/apps/catalog/models/test_taxonomy.py b/backend/metagenedb/apps/catalog/models/test_taxonomy.py
index b0d429539aab068b86a9405d87e310c2c371f8bb..16e6d47c7316125521c39ff2b0ee47c2ee8cb25f 100644
--- a/backend/metagenedb/apps/catalog/models/test_taxonomy.py
+++ b/backend/metagenedb/apps/catalog/models/test_taxonomy.py
@@ -60,3 +60,10 @@ class TestBuildHierarchy(APITestCase):
self.assertEqual(self.phylum.one_line_detailed_taxonomy, expected_str)
expected_str = "k__KINGDOM; p__PHYLUM; c__; o__; f__; g__; s__Species"
self.assertEqual(self.species.one_line_detailed_taxonomy, expected_str)
+
+ def test_csv(self):
+ expected = (
+ f"{self.species.tax_id},{self.species.name},"
+ f"{self.species.rank},{self.species.one_line_detailed_taxonomy}"
+ )
+ self.assertEqual(self.species.csv, expected)