def __init__(self, filetype, fn=None, pop=None):
self.builder = SibJoinBuilder(filetype, fn=fn, pop=pop)
self.d, self.threshold = self.builder.setupResults()
self.startTime = time.time()
self.run()
self.stopTime = time.time()
self.runTime = self.stopTime - self.startTime
class SibJoin:
def __init__(self, filetype, fn=None, pop=None):
self.builder = SibJoinBuilder(filetype, fn=fn, pop=pop)
self.d, self.threshold = self.builder.setupResults()
self.startTime = time.time()
self.run()
self.stopTime = time.time()
self.runTime = self.stopTime - self.startTime
def run(self):
allowable = SJGlobals.allowableJoins
d = self.d
clusters = SJGlobals.clusters
individuals = SJGlobals.individuals
nIndvs = SJGlobals.nIndvs
for t in range(self.threshold + 1):
largestFirst = []
for i in range(nIndvs):
for j in range(i + 1, nIndvs):
if d[i][j] == t and allowable[i][j]:
ind0 = individuals[i]
ind1 = individuals[j]
# Join full-siblings first
vFS = JoinTests.isValidFSWithHS(ind0, ind1)
if vFS[0]:
clusters.joinFS(ind0.fsCluster, ind1.fsCluster,\
vFS[1])
if len(vFS) == 3:
for fsJoin in vFS[2]:
clusters.join(fsJoin[1], fsJoin[0], fs=True)
# Construct half-sibs. Prefer larger families first
else:
for k in range(2):
for l in range(2):
cSize = min(\
len(clusters.hsClusters[ind0.hsClusters[k]].individuals),\
len(clusters.hsClusters[ind1.hsClusters[l]])\
)
largestFirst.append([cSize, i, j, k, l])
largestFirst.sort(reverse=True)
for fam in largestFirst:
# Don't merge families which have already been merged or are
# invalid
if d[fam[1]][fam[2]] == -1 and allowable[fam[1]][fam[2]]:
continue
ind0 = individuals[fam[1]]
ind1 = individuals[fam[2]]
clusterID0 = ind0.hsClusters[fam[3]]
clusterID1 = ind1.hsClusters[fam[4]]
vHS = JoinTests.isValidHS(clusterID0, clusterID1)
if vHS[0]:
tmp = sorted([clusterID0, clusterID1])
clusters.join(tmp[0], tmp[1])
for fsJoin in vHS[1]:
# fsJoin[1] is actually the smaller index
clusters.join(fsJoin[1], fsJoin[0], fs=True)
def getClusterings(self):
return SJGlobals.clusters.sortMaternalPaternal()
def getClusteringsLegacy(self):
tmpMHS, tmpPHS = SJGlobals.clusters.sortMaternalPaternal()
mHS = []
pHS = []
for cluster in tmpMHS:
mHS.append([ind.index for ind in cluster.individuals])
for cluster in tmpPHS:
pHS.append([ind.index for ind in cluster.individuals])
return mHS, pHS