Merge pull request #27 from ga4gh/sra_logical_ids_ID5

Sra logical ids id5

fasp/loc/DRSClient.py

@@ -1,5 +1,6 @@
 import requests
 import json
+#from builtins import None
 
 class DRSClient:
 	'''Basic DRS functions, no bundle handling'''    
@@ -67,5 +68,16 @@ def getAccessURL(self, object_id, access_id=None):
 			print (response.content)
 			return None
 
+
+	def getAccessURLRegion(self, object_id, region):
+		''' get an access url for the object in the specified region'''
+		access_methods = self.getObject(object_id)['access_methods']
+		am = next((sub for sub in access_methods if sub['region'] == 's3.us-east-1'), None)
+		if am == None:
+			print ('object not in region {}'.format(region))
+			return None
+		return self.getAccessURL(object_id, am['access_id'])
+
+
 	def getHeaders(self): 
 		return {'Authorization' : 'Bearer {0}'.format(self.access_token) }

fasp/loc/__init__.py

@@ -5,9 +5,9 @@
 from fasp.loc.gen3drsclient import anvilDRSClient
 from fasp.loc.gen3drsclient import kfDRSClient
 
-from fasp.loc.sbdrsclient import sbcgcDRSClient, cavaticaDRSClient, SBDRSClient
+from fasp.loc.sbdrsclient import sbcgcDRSClient, cavaticaDRSClient, sbbdcDRSClient, SBDRSClient
 
-from fasp.loc.sdl_drsclient import sdlDRSClient
+from fasp.loc.sdl_drsclient import sdlDRSClient, SRADRSClient
 
 from fasp.loc.ga4gh_registry_client import GA4GHRegistryClient
 

fasp/loc/drs_metaresolver.py

@@ -4,7 +4,7 @@
 
 from fasp.loc import crdcDRSClient, bdcDRSClient, Gen3DRSClient, anvilDRSClient
 from fasp.loc import sdlDRSClient
-from fasp.loc import sbcgcDRSClient, cavaticaDRSClient
+from fasp.loc import sbcgcDRSClient, cavaticaDRSClient, sbbdcDRSClient
 from fasp.loc import DRSClient
 from fasp.loc import GA4GHRegistryClient
 
@@ -23,7 +23,8 @@ def __init__(self, debug=False, getReg=True):
 			"anv": anvilDRSClient('~/.keys/anvil_credentials.json', '', 'gs'),
 			"insdc": sdlDRSClient('~/.keys/prj_11218_D17199.ngc'),
 			"sbcgc": sbcgcDRSClient('~/.keys/sevenbridges_keys.json','s3'),
-			"sbcav": cavaticaDRSClient('~/.keys/sevenbridges_keys.json','s3'),
+			"sbcav": cavaticaDRSClient('~/.keys/sevenbridges_keys.json','gs'),
+			'sbbdc' : sbbdcDRSClient('~/.keys/sevenbridges_keys.json', 's3'),
 			"srapub": DRSClient('https://locate.ncbi.nlm.nih.gov', debug=False)
 		}
 		self.registeredClients = []
@@ -35,7 +36,7 @@ def __init__(self, debug=False, getReg=True):
 	def getObject(self, colonPrefixedID):
 		client, id = self.getClient(colonPrefixedID)
 		if client != None:
-			print('sending id {} to: {}'.format(id, client.__class__.__name__))
+			if self.debug: print('sending id {} to: {}'.format(id, client.__class__.__name__))
 			return client.getObject(id)
 		else:
 			return "prefix unrecognized"

fasp/loc/sbdrsclient.py

@@ -40,18 +40,24 @@ def getHeaders(self):
 
 
 class sbcgcDRSClient(SBDRSClient):
-    '''client for Seven Bridges Cancer Genomics Cloud DRS server'''
+    '''client for Cancer Genomics Cloud Seven Bridges DRS server'''
 
     # Mostly done by the SBDRSClient, this just deals with url and end point specifics
     def __init__(self, api_key_path, access_id, debug=False):
     	super().__init__('https://cgc-ga4gh-api.sbgenomics.com', api_key_path, 'cgc', access_id, debug=debug)
 
 class cavaticaDRSClient(SBDRSClient):
-    '''client for Cavatica DRS Server'''    
+    '''client for Cavatica Seven Bridges DRS Server'''    
     # init mostly done by the SBDRSClient, this just deals with url and end point specifics
     def __init__(self, api_key_path, access_id, debug=False):
     	super().__init__('https://cavatica-ga4gh-api.sbgenomics.com', api_key_path, 'cavatica', access_id, debug=debug)
 
+class sbbdcDRSClient(SBDRSClient):
+    '''client for BioDataCatalyst Seven Bridges DRS Server'''    
+    # init mostly done by the SBDRSClient, this just deals with url and end point specifics
+    def __init__(self, api_key_path, access_id, debug=False):
+    	super().__init__('https://ga4gh-api.sb.biodatacatalyst.nhlbi.nih.gov', api_key_path, 'bdc', access_id, debug=debug)
+
 
 
 if __name__ == "__main__":

fasp/loc/sdl_drsclient.py

@@ -9,6 +9,22 @@
 
 from fasp.loc import DRSClient
 
+class SRADRSClient(DRSClient):
+	'''SRA DRS client with ability to convert SRA accessions to DRS ids'''
+
+	def __init__(self, api_url_base, access_id=None, debug=False, public=False):
+		super().__init__(api_url_base, access_id, debug, public)
+
+
+	def acc2drs(self, accession, verbose=False):
+		''' get a drs id for an SRA accession id'''
+		url = '{}/idx/v1/{}'.format(self.api_url_base, accession)
+		if verbose: print(url)
+		response = requests.get(url)
+		if verbose: print(response)
+		idxResp = json.loads(response.content)
+		return idxResp
+
 class sdlDRSClient(DRSClient):
 
 	def __init__(self, ngc_file, debug=False):

fasp/search/discovery_search_client.py

@@ -88,12 +88,19 @@ def listTableInfo(self, table, verbose=False):
 			print(json.dumps(info, indent=3))
 		return info
 
-	def listTableColumns(self, table, descriptions=False):
+	def listTableColumns(self, table, descriptions=False, enums=False):
 		''' List the columns in a table. More compact and practical for many purposes compared with listTableInfo '''
 		schema = self.listTableInfo(table)
 		for c, v in schema['data_model']['properties'].items():
 			print (c)
-			if descriptions: print (v['$comment'])
+			if descriptions:
+				if 'description' in v: print (v['description'])
+				if '$comment' in v: print (v['$comment'])
+			if enums:
+				if 'oneOf' in v:
+					for c in v['oneOf']:
+						print ('\t\t{}'.format(c['const']))
+				if '$comment' in v: print (v['$comment'])
 			print('_______________________________________')
 
 	def getMappingTemplate(self, table, propList=None):
@@ -142,6 +149,7 @@ def runQuery(self, query, returnType=None):
 				 headers=self.headers, data = query2)
 			else:
 				response = requests.request("GET", next_url)
+			if self.debug: print(response.content)
 			result = (response.json())
 			if self.debug:
 				pprint.pprint(result)
@@ -162,6 +170,9 @@ def runQuery(self, query, returnType=None):
 		else:
 			return resultRows
 
+	def query2Frame(self, query):
+		return self.runQuery(query, returnType='dataframe')
+
 
 def usage():
 	print (sys.argv[0] +' -l listTables -c listCatalog -t tableInfo -r registeredServices')