i_c = 1.0 / (2 * g * u * i_N)
i_m = 0.1
estates = 10
name = "ILS.test"
seqs = ["example_data.fa"]
# Creating an isolation model with three species is done by merging species
# pairwise.
modelILS = build_epoch_seperated_model(
3,
# we start with a,b,c.
[
# merge a and b to '0', c continues as '1'
[0, 0, 1],
# and now '0' and '1' is merged to a new '0'
[0, 0]
],
[1, estates, estates])
nstates = len(modelILS.tree_map)
names = ["bonobo", "pantro2", "hg18"]
init_ILS = (1.0e6 * u, 4.5e6 * u, i_c, i_r)
all_obs = []
total_L = 0
for seq in seqs:
# each sequence creates a new column map extended with any new columns that
# might be seen.
obs, colmap = readObservations(seq, names, COL_MAP)
COL_MAP = colmap
def runILSctmc(seqs, **args):
global COL_MAP
NeRef = args["NeRef"]
g = args["g"]
u = args["u"]
i_r = args["r"]
i_N1 = args["N1"]
i_N2 = args["N12"]
i_N3 = args["N123"]
i_c1 = 1.0/(2*g*u*i_N1)
i_c2 = 1.0/(2*g*u*i_N2)
i_c3 = 1.0/(2*g*u*i_N3)
i_t1 = u * args["T12"] / g / (2*NeRef)
i_t2 = u * args["T123"] / g / (2*NeRef)
startvalues = dict( [ (name,eval(name)) for name in ['i_t1', 'i_t2', 'i_c1', 'i_c2', 'i_c3', 'i_r'] ] )
# # generation time and mutation rate
# g = 20
# u = 1e-9
#
# # initial values for recombination rate, population size/coalescence rate and
# # migration rate.
# i_r = 0.4
# i_N = 100e3
# i_c = 1.0/(2*g*u*i_N)
# i_t1 = 3.7e6*u
# i_t2 = 5.95e6*u
estates = 4
print >>sys.stderr, 'running with %d epoch states' % estates
# inst = int(sys.argv[1])
#
# seqs = ["data/x_ils_%i.fa" % inst]
modelILS = build_epoch_seperated_model(3, [[0,0,1], [0,0]], [1,estates,estates])
nstates = len(modelILS.tree_map)
names = ["'1'", "'2'", "'0'"]
#names = ["'0'", "'2'", "'1'"]
all_obs = []
forwarders = []
COL_MAP = dict((v,i) for i,v in enumerate(product('ACGT', repeat=3)))
for seq in seqs:
# each sequence creates a new column map extended with any new columns that
# might be seen.
obs, colmap = readObservations(seq, names, COL_MAP)
all_obs.append(obs)
# print len(COL_MAP)
doEstimate = False
L = None
estimates = list()
if doEstimate:
for obs in all_obs:
print 'next obs:'
ffd, foutname = tempfile.mkstemp()
print ' temp fd/name:', ffd, foutname
fout = os.fdopen(ffd, 'w')
L = len(obs)
for j in xrange(L-1):
o = obs[j]
print >>fout, o,
print >>fout, obs[j]
fout.close()
print ' written, creating forwarder'
f = Forwarder.fromSequence(seqFilename = foutname, alphabetSize = len(COL_MAP))
#f = Forwarder(seqFilename = foutname, nStates = len(modelILS.tree_map), nObservables = len(COL_MAP))
print ' - done.'
forwarders.append(f)
os.system("rm %s" % foutname)
L, estimates = estimate_ILS(modelILS, forwarders, i_t1, i_t2, i_c1, i_c2, i_c3, i_r, outfile="/dev/null")
i_t1, i_t2, i_c1, i_c2, i_c3, i_r = estimates
estimates = dict( [ (name,eval(name)) for name in ['i_t1', 'i_t2', 'i_c1', 'i_c2', 'i_c3', 'i_r'] ] )
#print 'Estimates:'
#print "\t".join(map(str, [L] + est))
if "hook" in args:
args["hook"].run(modelILS, COL_MAP, all_obs, [i_c1, i_c2, i_c3], [i_r]*3, [0]*3, [0, i_t1, i_t2])
return L, estimates, startvalues
def main():
usage = """%prog [options] <forwarder dir>
This program calculates the posterior state probabilities of an isolation
model with two species and uniform coalescence/recombination rate."""
parser = OptionParser(usage=usage, version="%prog 1.0")
parser.add_option("-o",
"--out",
dest="outfile",
type="string",
default="/dev/stdout",
help="Output file for the estimate (/dev/stdout)")
parser.add_option(
"-n",
"--seq-name",
dest="seq_name",
type="string",
default="sequence",
help="Name of the sequence. Used in the output for tabix indexing")
parser.add_option(
"-p",
"--first-pos",
dest="first_pos",
type="int",
default=0,
help="Position in the sequence where the first element is."
"Used in the output for tabix indexing. Default 1.")
optimized_params = [
('T', 'split time'),
('C', 'Coalescence rate'),
('R', 'recombination rate'),
]
for (cname, desc) in optimized_params:
parser.add_option("-%s" % cname,
dest=cname,
type="float",
help="Model parameter %s" % desc)
parser.add_option(
"--intervals",
dest="intervals",
type="int",
default=10,
help="Number of sub intervals used to discretize the time (10)")
parser.add_option("--no-header",
dest="no_header",
action="store_true",
default=False,
help="Do not include the header in the output")
(options, args) = parser.parse_args()
if len(args) != 1:
parser.error("Needs a preprocessed sequence to work on")
split_time = options.T
coal_rate = options.C
rec_rate = options.R
if split_time is None:
print 'You must specify the split time.'
sys.exit(2)
if coal_rate is None:
print 'You must specify the coalescence rate.'
sys.exit(2)
if rec_rate is None:
print 'You must specify the recombination rate.'
sys.exit(2)
intervals = options.intervals
model = build_epoch_seperated_model(2, [[0, 0]], [1, intervals])
mpi, mT, mE, time_breaks = get_model_matrices(model, coal_rate, rec_rate,
split_time)
pi, T, E = zipHMM_prepare_matrices(mpi, mT, mE)
with open(options.outfile, 'w') as fout:
if not options.no_header:
print >> fout, '## Posterior probabilities for isolation model.'
print >> fout, '# intervals =', intervals
print >> fout, '# T =', split_time
print >> fout, '# C =', coal_rate
print >> fout, '# R =', rec_rate
print >> fout, '# time_breaks =', ' '.join(map(str, time_breaks))
try:
for ziphmmdir in args:
seqfile = os.path.join(ziphmmdir, 'original_sequence')
_, pdTable = posteriorDecoding(seqfile, pi, T, E)
print_matrix(pdTable, options.seq_name, options.first_pos,
fout)
except IOError as e:
if e.errno == errno.EPIPE:
sys.exit(0) # the pipe died, this is probably because it is in
# a shell pipe where we have stopped reading
else:
raise e
i_c = 1.0/(2*g*u*i_N)
i_m = 0.1
estates = 10
name = "ILS.test"
seqs = ["example_data.fa"]
# Creating an isolation model with three species is done by merging species
# pairwise.
modelILS = build_epoch_seperated_model(
3,
# we start with a,b,c.
[
# merge a and b to '0', c continues as '1'
[0,0,1],
# and now '0' and '1' is merged to a new '0'
[0,0]
],
[1,estates,estates])
nstates = len(modelILS.tree_map)
names = ["bonobo", "pantro2", "hg18"]
init_ILS = (1.0e6*u, 4.5e6*u, i_c, i_r)
all_obs = []
total_L = 0
for seq in seqs:
# each sequence creates a new column map extended with any new columns that
# might be seen.
obs, colmap = readObservations(seq, names, COL_MAP)
COL_MAP = colmap
def main():
usage = """%prog [options] <forwarder dirs>
This program estimates the parameters of an isolation model with two species
and uniform coalescence/recombination rate."""
parser = OptionParser(usage=usage, version="%prog 1.0")
parser.add_option("-o",
"--out",
dest="outfile",
type="string",
default="/dev/stdout",
help="Output file for the estimate (/dev/stdout)")
parser.add_option(
"--tmpfile",
dest="tmpfile",
type="string",
default="/dev/null",
help="Log for all points estimated in the optimization (/dev/null)")
optimized_params = [
('splittime', 'split time', 1e6),
('Ne', 'effective population size', 20e3),
('recomb', 'recombination rate', 0.1),
]
for (cname, desc, default) in optimized_params:
parser.add_option("--%s" % cname,
dest=cname,
type="float",
default=default,
help="Initial guess at the %s (%g)" %
(desc, default))
fixed_params = [
('mu', 'mutation rate', 1e-9),
('g', 'generation time', 20),
]
for (cname, desc, default) in fixed_params:
parser.add_option("--%s" % cname,
dest=cname,
type="float",
default=default,
help="Value of the %s (%g)" % (desc, default))
parser.add_option(
"--intervals",
dest="intervals",
type="int",
default=10,
help="Number of sub intervals used to discretize the time (10)")
parser.add_option("--header",
dest="include_header",
action="store_true",
default=False,
help="Include a header on the output")
parser.add_option("-v",
"--verbose",
dest="verbose",
action="store_true",
default=False,
help="Print help")
(options, args) = parser.parse_args()
if len(args) < 1:
parser.error("Needs at least one preprocessed sequence to work on")
if not options.verbose:
log = lambda s: None
logu = lambda s: None
else:
logu = log_unfinished_line
log = log_finished_line
logu("Loading forwarders...")
forwarders = [Forwarder.fromDirectory(dir) for dir in args]
log("done")
logu("Constructing model...")
intervals = options.intervals
modelI = build_epoch_seperated_model(2, [[0, 0]], [1, intervals])
log("done")
mu = options.mu
g = options.g
T = options.splittime * mu
C = 1.0 / (g * mu * 2 * options.Ne)
R = options.recomb
with open(options.tmpfile, 'w') as tmpfile:
L, est = estimate_I(modelI, forwarders, T, C, R, outfile=tmpfile)
vals = "\t".join(map(str, est))
with open(options.outfile, 'w') as outfile:
if options.include_header:
print >> outfile, 'logL\tT\tC\tR'
print >> outfile, "%f\t%s" % (L, vals)
def main():
usage="""%prog [options] <forwarder dirs>
This program estimates the parameters of an isolation model with two species
and uniform coalescence/recombination rate."""
parser = OptionParser(usage=usage, version="%prog 1.0")
parser.add_option("-o", "--out",
dest="outfile",
type="string",
default="/dev/stdout",
help="Output file for the estimate (/dev/stdout)")
parser.add_option("--tmpfile",
dest="tmpfile",
type="string",
default="/dev/null",
help="Log for all points estimated in the optimization (/dev/null)")
optimized_params = [
('splittime', 'split time', 1e6),
('Ne', 'effective population size', 20e3),
('recomb', 'recombination rate', 0.1),
]
for (cname, desc, default) in optimized_params:
parser.add_option("--%s" % cname,
dest=cname,
type="float",
default=default,
help="Initial guess at the %s (%g)" % (desc, default))
fixed_params = [
('mu', 'mutation rate', 1e-9),
('g', 'generation time', 20),
]
for (cname, desc, default) in fixed_params:
parser.add_option("--%s" % cname,
dest=cname,
type="float",
default=default,
help="Value of the %s (%g)" % (desc, default))
parser.add_option("--intervals",
dest="intervals",
type="int",
default=10,
help="Number of sub intervals used to discretize the time (10)")
parser.add_option("--header",
dest="include_header",
action="store_true",
default=False,
help="Include a header on the output")
parser.add_option("-v", "--verbose",
dest="verbose",
action="store_true",
default=False,
help="Print help")
(options, args) = parser.parse_args()
if len(args) < 1:
parser.error("Needs at least one preprocessed sequence to work on")
if not options.verbose:
log = lambda s: None
logu = lambda s: None
else:
logu = log_unfinished_line
log = log_finished_line
logu("Loading forwarders...")
forwarders = [Forwarder.fromDirectory(dir) for dir in args]
log("done")
logu("Constructing model...")
intervals = options.intervals
modelI = build_epoch_seperated_model(2, [[0,0]], [1,intervals])
log("done")
mu = options.mu
g = options.g
T = options.splittime * mu
C = 1.0/(g*mu*2*options.Ne)
R = options.recomb
with open(options.tmpfile, 'w') as tmpfile:
L, est = estimate_I(modelI, forwarders, T, C, R, outfile=tmpfile)
vals = "\t".join(map(str,est))
with open(options.outfile, 'w') as outfile:
if options.include_header:
print >>outfile, 'logL\tT\tC\tR'
print >>outfile, "%f\t%s" % (L,vals)
def main():
usage="""%prog [options] <forwarder dir>
This program calculates the posterior state probabilities of an isolation
model with two species and uniform coalescence/recombination rate."""
parser = OptionParser(usage=usage, version="%prog 1.0")
parser.add_option("-o", "--out",
dest="outfile",
type="string",
default="/dev/stdout",
help="Output file for the estimate (/dev/stdout)")
parser.add_option("-n", "--seq-name",
dest="seq_name",
type="string",
default="sequence",
help="Name of the sequence. Used in the output for tabix indexing")
parser.add_option("-p", "--first-pos",
dest="first_pos",
type="int",
default=0,
help="Position in the sequence where the first element is."
"Used in the output for tabix indexing. Default 1.")
optimized_params = [
('T', 'split time'),
('C', 'Coalescence rate'),
('R', 'recombination rate'),
]
for (cname, desc) in optimized_params:
parser.add_option("-%s" % cname,
dest=cname,
type="float",
help="Model parameter %s" % desc)
parser.add_option("--intervals",
dest="intervals",
type="int",
default=10,
help="Number of sub intervals used to discretize the time (10)")
parser.add_option("--no-header",
dest="no_header",
action="store_true",
default=False,
help="Do not include the header in the output")
(options, args) = parser.parse_args()
if len(args) != 1:
parser.error("Needs a preprocessed sequence to work on")
split_time = options.T
coal_rate = options.C
rec_rate = options.R
if split_time is None:
print 'You must specify the split time.'
sys.exit(2)
if coal_rate is None:
print 'You must specify the coalescence rate.'
sys.exit(2)
if rec_rate is None:
print 'You must specify the recombination rate.'
sys.exit(2)
intervals = options.intervals
model = build_epoch_seperated_model(2, [[0,0]], [1,intervals])
mpi, mT, mE, time_breaks = get_model_matrices(model, coal_rate, rec_rate, split_time)
pi, T, E = zipHMM_prepare_matrices(mpi, mT, mE)
with open(options.outfile,'w') as fout:
if not options.no_header:
print >> fout, '## Posterior probabilities for isolation model.'
print >> fout, '# intervals =', intervals
print >> fout, '# T =', split_time
print >> fout, '# C =', coal_rate
print >> fout, '# R =', rec_rate
print >> fout, '# time_breaks =', ' '.join(map(str, time_breaks))
try:
for ziphmmdir in args:
seqfile = os.path.join(ziphmmdir, 'original_sequence')
_, pdTable = posteriorDecoding(seqfile, pi, T, E)
print_matrix(pdTable, options.seq_name, options.first_pos, fout)
except IOError as e:
if e.errno == errno.EPIPE:
sys.exit(0) # the pipe died, this is probably because it is in
# a shell pipe where we have stopped reading
else:
raise e