def test_getRep(self):
'''getRep should return the desired number of sequences '''
aaUsage = tcrRepSim.readAaUsage(self.aaUsageFile)
codonUsage = tcrRepSim.readCodonUsage(self.codonUsageFile)
seqs = tcrRepSim.getUniqSeqRep(10, aaUsage[9], codonUsage)
tcrRepSim.getRep( seqs, 50, None )
total = sum([s.count for s in seqs])
self.assertTrue( total == 50 )
def test_getUniqSeqRep(self):
'''getUniqSeqRep should return the desired number of uniq sequences '''
aaUsage = tcrRepSim.readAaUsage(self.aaUsageFile)
codonUsage = tcrRepSim.readCodonUsage(self.codonUsageFile)
seqs = tcrRepSim.getUniqSeqRep(100, aaUsage[9], codonUsage)
self.assertTrue( len(seqs) == 100)
visited = []
for s in seqs:
self.assertTrue( s.nuc not in visited )
visited .append(s.nuc)
def test_sampling(self):
'''sampling should return the desired number of sequences '''
aaUsage = tcrRepSim.readAaUsage(self.aaUsageFile)
codonUsage = tcrRepSim.readCodonUsage(self.codonUsageFile)
seqs = tcrRepSim.getUniqSeqRep(10, aaUsage[9], codonUsage)
tcrRepSim.getRep( seqs, 50, None )
samseqs = tcrRepSim.sampling( seqs, 20 )
total = sum([s.count for s in samseqs])
self.assertTrue( len(samseqs) <= 10 )
self.assertTrue( total == 20 )
def test_readAaUsage_goodInput(self):
'''readAaUsage should return correct known usage'''
len2aaUsage = tcrRepSim.readAaUsage( self.goodEx )
lens = [2, 4]
knownUsage = {2:[{'C':9999.0/10000, 'A':1.0/10000}, {'F':3.0/3}], 4:[ {'C':100.0/100}, {'A':79.0/100, 'R': 1.0/100, 'S':20.0/100}, {'A':500.0/1000, 'C':200.0/1000, 'G':300.0/1000},{'F':1}] }
for l in lens:
self.assertTrue( l in len2aaUsage )
self.assertTrue( len( len2aaUsage[l] ) == l )
aaUsage = len2aaUsage[l]
known = knownUsage[l]
for i in xrange(l):
s = aaUsage[i]
freqs = known[i]
a2c = {}
for a in s:
if a in a2c:
a2c[a] += 1
else:
a2c[a] = 1
for a in a2c:
a2c[a] = float( a2c[a] )/len(s)
self.assertTrue( a2c[a] == freqs[a] )
