def importEdges(filename=None):
"""
import edges, one pair per line
Edge property: - no direction, -> left to right, <- right to left
"""
filename = filename or r'C:\work\caida\Dropbox\precon\engineering\Contents\GBM_BN-Massaged.csv'
# add a network
n = Network.findOne({'name': "GBM Predicted Tumor Network"}) or Network()
n.name = "GBM Predicted Tumor Network"
n.group = "predicted"
n.owner = 'precon'
n.save()
col = mongo.db()['entity']
count = 0
now = time.time()
ec = mongo.getCollection('edge')
with open(filename) as f:
content = f.read()
for line in content.split('\n'):
doc = {}
line = line.strip()
pos = line.find("///")
if pos > 0:
line = line[0:pos].strip()
doc['comment'] = line[pos:]
items = line.split("->")
if (len(items) == 1): items = line.split("<-")
if (len(items) == 1): items = line.split(" - ")
if (len(items) == 1):
error("Ignore invalid line: [%s]" % line)
continue
count += 1
if (count < 8378): continue
tmp = []
tmp.append(items[0].lower().strip())
tmp.append(items[1].lower().strip())
entities = ['', '']
print "!!! %d " % (col.find({'symbol': {'$in': tmp}}).count())
for r in col.find({'symbol': {'$in': tmp}}):
if (r['symbol'] == tmp[0]):
entities[0] = Entity(r)
if (r['symbol'] == tmp[1]):
entities[1] = Entity(r)
if (len(entities) != 2):
raise "Invalid nodes %s, continue" % entities
node1 = Node.findOne({'network': n._id, 'entity': entities[0]._id})
if not node1:
node1 = Node({'network': n._id}, entities[0])
node1.save()
node2 = Node.findOne({'network': n._id, 'entity': entities[1]._id})
if not node2:
node2 = Node({'network': n._id}, entities[1])
node2.save()
con = Connection()
con._id = "conn_%s_%s" % (tmp[0], tmp[1])
con.nodes = [node1._id, node2._id]
con.entities = [entities[0]._id, entities[1]._id]
con.type = 'predicted'
con.network = n._id
con.label = ""
con.save()
print "Saving connection %d %s" % (count, con._id)
finish = time.time()
print "Imported %d edges, time elpased %d seconds" % (count, finish - now)
def newNode(self, entity): node = Node(entity=entity) return node
def importEdges(filename=None):
"""
import edges, one pair per line
Edge property: - no direction, -> left to right, <- right to left
"""
filename = filename or r'C:\work\caida\Dropbox\precon\engineering\Contents\GBM_BN-Massaged.csv'
# add a network
n = Network.findOne({'name': "GBM Predicted Tumor Network"}) or Network()
n.name = "GBM Predicted Tumor Network"
n.group = "predicted"
n.owner = 'precon'
n.save()
col = mongo.db()['entity']
count = 0
now = time.time()
ec = mongo.getCollection('edge')
with open(filename) as f:
content = f.read()
for line in content.split('\n'):
doc = {}
line = line.strip()
pos = line.find("///")
if pos>0:
line = line[0:pos].strip()
doc['comment'] = line[pos:]
items = line.split("->")
if (len(items) == 1): items = line.split("<-")
if (len(items) == 1): items = line.split(" - ")
if (len(items) == 1):
error("Ignore invalid line: [%s]" %line)
continue
count+=1
if(count<8378): continue
tmp =[]
tmp.append( items[0].lower().strip())
tmp.append( items[1].lower().strip())
entities = ['','']
print "!!! %d " %( col.find({'symbol': {'$in': tmp } }).count() )
for r in col.find({'symbol': {'$in': tmp } }):
if(r['symbol'] == tmp[0]):
entities[0] = Entity(r)
if(r['symbol'] == tmp[1]):
entities[1] = Entity(r)
if(len(entities)!=2 ):
raise "Invalid nodes %s, continue" % entities
node1 = Node.findOne({'network': n._id, 'entity': entities[0]._id})
if not node1:
node1 = Node({'network':n._id}, entities[0])
node1.save()
node2 = Node.findOne({'network': n._id, 'entity': entities[1]._id})
if not node2:
node2 = Node({'network':n._id}, entities[1])
node2.save()
con = Connection()
con._id = "conn_%s_%s" %(tmp[0], tmp[1])
con.nodes = [node1._id, node2._id]
con.entities = [ entities[0]._id, entities[1]._id ]
con.type = 'predicted'
con.network = n._id
con.label = ""
con.save()
print "Saving connection %d %s" % (count, con._id)
finish = time.time()
print "Imported %d edges, time elpased %d seconds" %(count, finish - now)
def parseInteractions(dom,network):
"""
<interaction id="678614" imexId="IM-12113-2">
<names>
<shortLabel>pf1130-pf1128-2</shortLabel>
<fullName>Affinity purification by hydroxy apatite chromatography</fullName>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="EBI-2507294" dbAc="MI:0469" db="intact"/>
<secondaryRef refType="imex source" id="MI:0469" dbAc="MI:0488" db="psi-mi"/>
<secondaryRef refTypeAc="MI:0662" refType="imex-primary" id="IM-12113-2" dbAc="MI:0670" db="imex"/>
</xref>
<experimentList>
<experimentRef>678502</experimentRef>
</experimentList>
<participantList>
<participant id="678615">
<names>
<shortLabel>n/a</shortLabel>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="EBI-2507299" dbAc="MI:0469" db="intact"/>
</xref>
<interactorRef>678518</interactorRef>
<biologicalRole>
<names>
<shortLabel>unspecified role</shortLabel>
<fullName>unspecified role</fullName>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="MI:0499" dbAc="MI:0488" db="psi-mi"/>
<secondaryRef refTypeAc="MI:0356" refType="identity" id="EBI-77781" dbAc="MI:0469" db="intact"/>
<secondaryRef refTypeAc="MI:0358" refType="primary-reference" id="14755292" dbAc="MI:0446" db="pubmed"/>
</xref>
</biologicalRole>
<experimentalRoleList>
<experimentalRole>
<names>
<shortLabel>neutral component</shortLabel>
<fullName>neutral component</fullName>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="MI:0497" dbAc="MI:0488" db="psi-mi"/>
<secondaryRef refTypeAc="MI:0356" refType="identity" id="EBI-55" dbAc="MI:0469" db="intact"/>
<secondaryRef refTypeAc="MI:0358" refType="primary-reference" id="14755292" dbAc="MI:0446" db="pubmed"/>
</xref>
</experimentalRole>
</experimentalRoleList>
</participant>
"""
ems = dom.getElementsByTagName("interaction")
connections = []
nodeCount = 0
for em in ems:
#interactions.append(interaction)
participants = em.getElementsByTagName("participant")
con_nodes = []
for p in participants:
node = Node()
node.network = network._id
nodeCount+=1
node._id = "node%s_%d" %(node.network[4:], nodeCount)
node.role = getText(p, "biologicalRole", "names", "shortLabel")
#node.refs['intact'] = p.getAttribute("id")
setPrimaryRef(p, node)
ref = getText(p, 'interactorRef')
if ref: node.refs['entity'] = ref # temporary
con_nodes.append(node)
connection = Connection()
connection._id = "conn_intact_%s" % em.getAttribute("id")
connection.refs={}
setPrimaryRef(em, connection)
connection.type=getText( em.getElementsByTagName("interactionType")[0], 'shortLabel')
connection.label = getText(em,'shortLabel')
connection.network = network._id
connection.nodes = con_nodes
connections.append(connection)
network.connections = connections
def parseInteractions(dom, network):
"""
<interaction id="678614" imexId="IM-12113-2">
<names>
<shortLabel>pf1130-pf1128-2</shortLabel>
<fullName>Affinity purification by hydroxy apatite chromatography</fullName>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="EBI-2507294" dbAc="MI:0469" db="intact"/>
<secondaryRef refType="imex source" id="MI:0469" dbAc="MI:0488" db="psi-mi"/>
<secondaryRef refTypeAc="MI:0662" refType="imex-primary" id="IM-12113-2" dbAc="MI:0670" db="imex"/>
</xref>
<experimentList>
<experimentRef>678502</experimentRef>
</experimentList>
<participantList>
<participant id="678615">
<names>
<shortLabel>n/a</shortLabel>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="EBI-2507299" dbAc="MI:0469" db="intact"/>
</xref>
<interactorRef>678518</interactorRef>
<biologicalRole>
<names>
<shortLabel>unspecified role</shortLabel>
<fullName>unspecified role</fullName>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="MI:0499" dbAc="MI:0488" db="psi-mi"/>
<secondaryRef refTypeAc="MI:0356" refType="identity" id="EBI-77781" dbAc="MI:0469" db="intact"/>
<secondaryRef refTypeAc="MI:0358" refType="primary-reference" id="14755292" dbAc="MI:0446" db="pubmed"/>
</xref>
</biologicalRole>
<experimentalRoleList>
<experimentalRole>
<names>
<shortLabel>neutral component</shortLabel>
<fullName>neutral component</fullName>
</names>
<xref>
<primaryRef refTypeAc="MI:0356" refType="identity" id="MI:0497" dbAc="MI:0488" db="psi-mi"/>
<secondaryRef refTypeAc="MI:0356" refType="identity" id="EBI-55" dbAc="MI:0469" db="intact"/>
<secondaryRef refTypeAc="MI:0358" refType="primary-reference" id="14755292" dbAc="MI:0446" db="pubmed"/>
</xref>
</experimentalRole>
</experimentalRoleList>
</participant>
"""
ems = dom.getElementsByTagName("interaction")
connections = []
nodeCount = 0
for em in ems:
#interactions.append(interaction)
participants = em.getElementsByTagName("participant")
con_nodes = []
for p in participants:
node = Node()
node.network = network._id
nodeCount += 1
node._id = "node%s_%d" % (node.network[4:], nodeCount)
node.role = getText(p, "biologicalRole", "names", "shortLabel")
#node.refs['intact'] = p.getAttribute("id")
setPrimaryRef(p, node)
ref = getText(p, 'interactorRef')
if ref: node.refs['entity'] = ref # temporary
con_nodes.append(node)
connection = Connection()
connection._id = "conn_intact_%s" % em.getAttribute("id")
connection.refs = {}
setPrimaryRef(em, connection)
connection.type = getText(
em.getElementsByTagName("interactionType")[0], 'shortLabel')
connection.label = getText(em, 'shortLabel')
connection.network = network._id
connection.nodes = con_nodes
connections.append(connection)
network.connections = connections