iGEM as sequence source (#242)

* working version of #240

* update to new conventions

* update linkml library to pypi

Author: manulera

main.py

@@ -49,6 +49,7 @@
     OverlapExtensionPCRLigationSource,
     GatewaySource,
     AnnotationSource,
+    IGEMSource,
 )
 from fastapi.middleware.cors import CORSMiddleware
 from Bio.Restriction.Restriction import RestrictionBatch
@@ -474,6 +475,22 @@ async def get_from_repository_id_euroscarf(source: EuroscarfSource):
         repository_id_http_error_handler(exception, source)
 
 
+@router.post(
+    '/repository_id/igem',
+    response_model=create_model(
+        'IGEMResponse', sources=(list[IGEMSource], ...), sequences=(list[TextFileSequence], ...)
+    ),
+)
+async def get_from_repository_id_igem(source: IGEMSource):
+    try:
+        dseq = (await get_sequences_from_gb_file_url(source.sequence_file_url))[0]
+        return {'sequences': [format_sequence_genbank(dseq, source.output_name)], 'sources': [source]}
+    except HTTPError as exception:
+        repository_id_http_error_handler(exception, source)
+    except URLError as exception:
+        repository_id_url_error_handler(exception, source)
+
+
 @router.post(
     '/genome_coordinates',
     response_model=create_model(

poetry.lock

@@ -40,6 +40,7 @@
     GatewaySource as _GatewaySource,
     InFusionSource as _InFusionSource,
     AnnotationSource as _AnnotationSource,
+    IGEMSource as _IGEMSource,
 )
 from pydna.utils import shift_location as _shift_location
 from assembly2 import edge_representation2subfragment_representation, subfragment_representation2edge_representation
@@ -128,6 +129,15 @@ class EuroscarfSource(SourceCommonClass, _EuroscarfSource):
     pass
 
 
+class IGEMSource(SourceCommonClass, _IGEMSource):
+
+    @model_validator(mode='after')
+    def validate_repository_id(self):
+        file_name = self.sequence_file_url.split('/')[-1]
+        assert file_name.endswith('.gb'), 'The sequence file must be a GenBank file'
+        return self
+
+
 class GenomeCoordinatesSource(SourceCommonClass, _GenomeCoordinatesSource):
     pass
 

pydantic_models.py

@@ -17,7 +17,7 @@ pydna = {git = "https://github.com/BjornFJohansson/pydna", rev = "9d112d71534194
 requests = "^2.31.0"
 regex = "^2023.10.3"
 pydantic = "^2.7.1"
-shareyourcloning-linkml = "0.1.10a0"
+shareyourcloning-linkml = "0.1.11a0"
 pandas = "^2.2.3"
 openpyxl = "^3.1.5"
 

pyproject.toml

@@ -29,6 +29,7 @@
     SnapGenePlasmidSource,
     GatewaySource,
     AnnotationSource,
+    IGEMSource,
 )
 from pydna.dseqrecord import Dseqrecord
 import unittest
@@ -2714,5 +2715,41 @@ def test_get_gateway_sites(self):
         self.assertIn('attP1', payload)
 
 
+class IGEMSourceTest(unittest.TestCase):
+    good_url = 'https://raw.githubusercontent.com/manulera/annotated-igem-distribution/master/results/plasmids/1.gb'
+    no_gb_url = 'https://blah.com/1.txt'
+    wrong_url = (
+        'https://raw.githubusercontent.com/manulera/annotated-igem-distribution/master/results/plasmids/dummy.gb'
+    )
+
+    def test_igem(self):
+        source = IGEMSource(
+            id=0, repository_name='igem', repository_id='BBa_C0062-dummy', sequence_file_url=self.good_url
+        )
+        response = client.post('/repository_id/igem', json=source.model_dump())
+        payload = response.json()
+        self.assertEqual(response.status_code, 200)
+        self.assertEqual(payload['sources'][0]['repository_id'], 'BBa_C0062-dummy')
+        self.assertEqual(payload['sources'][0]['repository_name'], 'igem')
+
+    def test_errors(self):
+
+        # The repository_id does not start with the part_name, even if url is valid
+        source = IGEMSource(
+            id=0, repository_name='igem', repository_id='BBa_C0062-dummy', sequence_file_url=self.good_url
+        )
+
+        # The url is not a GenBank file
+        source_json = source.model_dump()
+        source_json['sequence_file_url'] = self.no_gb_url
+        response = client.post('/repository_id/igem', json=source_json)
+        self.assertEqual(response.status_code, 422)
+
+        # The url does not exist
+        source = IGEMSource(id=0, repository_name='igem', repository_id='dummy-test', sequence_file_url=self.wrong_url)
+        response = client.post('/repository_id/igem', json=source.model_dump())
+        self.assertEqual(response.status_code, 404)
+
+
 if __name__ == '__main__':
     unittest.main()