def getGenomeNeighborhoodsAndRoles(genomes, config):
cdmi_entity = CDMI_EntityAPI(config["cdmi_url"])
pegs = genomesToPegs(genomes)
# Get contigs
fidlocdict = cdmi_entity.get_relationship_IsLocatedIn(pegs, [], ["begin", "dir"], ["id"])
fids = getFieldFromRelationship(fidlocdict, "from_link", "rel")
begins = getFieldFromRelationship(fidlocdict, "begin", "rel")
dirs = getFieldFromRelationship(fidlocdict, "dir", "rel")
cids = getFieldFromRelationship(fidlocdict, "id", "to")
tuplist = []
for ii in range(len(cids)):
tuplist.append( (cids[ii], fids[ii], int(begins[ii]), dirs[ii]) )
# Sort by contig first, then by start location.
tuplist = sorted(tuplist, key=operator.itemgetter(0,2))
# Now lets get the role for all of these IDs
# Note that a single protein can have multiple roles.
roledict = cdmi_entity.get_relationship_HasFunctional(fids, [], [], ["id"])
fids = getFieldFromRelationship(roledict, "from_link", "rel")
roles = getFieldFromRelationship(roledict, "id", "to")
fidToRoles = {}
rolesToFids = {}
for ii in range(len(fids)):
if fids[ii] in fidToRoles:
fidToRoles[fids[ii]].append(roles[ii])
else:
fidToRoles[fids[ii]] = [ roles[ii] ]
if roles[ii] in rolesToFids:
rolesToFids[roles[ii]].append(fids[ii])
else:
rolesToFids[roles[ii]] = [ fids[ii] ]
return tuplist, fidToRoles
def fidsToRoles(fidlist, config):
''' Given a list of feature IDs return a dictionary from FID to the list of roles the encoding gene
performs and a dictionary from roles to the FIDs performing them.
@param fidlist List of feature IDs
@param config Dictionary of configuration variables
@return Dictionary keyed by feature ID of list of roles encoding gene performs, dictionary
keyed by role of list of feature IDs performing the role
'''
cdmi = CDMI_API(config["cdmi_url"])
cdmi_entity = CDMI_EntityAPI(config["cdmi_url"])
# Break the complete list into smaller sub-lists to avoid timeouts
start = 0
increment = 1000
end = start + increment
counter = len(fidlist)
fidsToRoles = {}
rolesToFids = {}
while counter > 0:
try:
roledict = cdmi_entity.get_relationship_HasFunctional(fidlist[start:end], [], [], ["id"])
except HTTPError as e:
if increment > 1:
increment = increment / 2
end = start + increment
sys.stderr.write("caught '%s' error, increment is now %d\n" %(e.reason, increment))
continue
flist = getFieldFromRelationship(roledict, "from_link", "rel")
rolelist = getFieldFromRelationship(roledict, "id", "to")
for ii in range(len(flist)):
# We have to use sets here because a bug(?) in get_relationship_HasFunctional allows multiple identical
# links between fids and roles.
# See for example what happens when you call it on g.9647.peg.2332
if flist[ii] in fidsToRoles:
fidsToRoles[flist[ii]].add(rolelist[ii])
else:
fidsToRoles[flist[ii]] = set([rolelist[ii]])
if rolelist[ii] in rolesToFids:
rolesToFids[rolelist[ii]].add(flist[ii])
else:
rolesToFids[rolelist[ii]] = set([flist[ii]])
# Move to next sub-list
start += increment
end += increment
if end >= len(fidlist):
end = len(fidlist)
counter -= increment
# Convert back to lists to not break other functions.
for f in fidsToRoles:
fidsToRoles[f] = list(fidsToRoles[f])
for r in rolesToFids:
rolesToFids[r] = list(rolesToFids[r])
return fidsToRoles, rolesToFids