def optimize_ordering(fwtour, clm, phase, cpus):
"""
Optimize the ordering of contigs by Genetic Algorithm (GA).
"""
from .chic import score_evaluate_M
# Prepare input files
tour_contigs = clm.active_contigs
tour_sizes = clm.active_sizes
tour_M = clm.M
tour = clm.tour
signs = clm.signs
oo = clm.oo
def callback(tour, gen, phase, oo):
fitness = tour.fitness if hasattr(tour, "fitness") else None
label = "GA{}-{}".format(phase, gen)
if fitness:
fitness = "{0}".format(fitness).split(",")[0].replace("(", "")
label += "-" + fitness
if gen % 20 == 0:
print_tour(fwtour, tour, label, tour_contigs, oo, signs=signs)
return tour
callbacki = partial(callback, phase=phase, oo=oo)
toolbox = GA_setup(tour)
toolbox.register("evaluate", score_evaluate_M,
tour_sizes=tour_sizes, tour_M=tour_M)
tour, tour_fitness = GA_run(toolbox, ngen=1000, npop=100, cpus=cpus,
callback=callbacki)
clm.tour = tour
return tour
def optimize_ordering(fwtour, clm, phase, cpus):
"""
Optimize the ordering of contigs by Genetic Algorithm (GA).
"""
from .chic import score_evaluate_M
# Prepare input files
tour_contigs = clm.active_contigs
tour_sizes = clm.active_sizes
tour_M = clm.M
tour = clm.tour
signs = clm.signs
oo = clm.oo
def callback(tour, gen, phase, oo):
fitness = tour.fitness if hasattr(tour, "fitness") else None
label = "GA{}-{}".format(phase, gen)
if fitness:
fitness = "{0}".format(fitness).split(",")[0].replace("(", "")
label += "-" + fitness
if gen % 20 == 0:
print_tour(fwtour, tour, label, tour_contigs, oo, signs=signs)
return tour
callbacki = partial(callback, phase=phase, oo=oo)
toolbox = GA_setup(tour)
toolbox.register("evaluate", score_evaluate_M,
tour_sizes=tour_sizes, tour_M=tour_M)
tour, tour_fitness = GA_run(toolbox, ngen=1000, npop=100, cpus=cpus,
callback=callbacki)
clm.tour = tour
return tour
def __init__(self, lgs, scaffolds, mapc, pivot, weights, sizes,
function=(lambda x: x.rank), linkage=min,
ngen=500, npop=100, cpus=8):
self.lgs = lgs
self.lengths = mapc.lengths
self.bins = mapc.bins
self.sizes = sizes
self.scaffolds = scaffolds
self.pivot = pivot
self.weights = weights
self.function = function
self.linkage = linkage
self.prepare_linkage_groups() # populate all data
signs = self.assign_orientation()
assert len(signs) == len(scaffolds)
scaffolds_oo = dict(zip(scaffolds, signs))
tour = self.assign_order()
tour = [(x, scaffolds_oo[x]) for x in tour]
i = 0
best_tour, best_fitness = None, None
while True: # Multiple EC rounds due to orientation fixes
logging.debug("Start EC round {0}".format(i))
scaffolds_oo = dict(tour)
scfs, tour, ww = self.prepare_ec(scaffolds, tour, weights)
toolbox = GA_setup(tour)
toolbox.register("evaluate", colinear_evaluate_multi,
scfs=scfs, weights=ww)
tour, fitness = GA_run(toolbox, ngen=ngen, npop=npop, cpus=cpus)
tour = [scaffolds[x] for x in tour]
tour = [(x, scaffolds_oo[x]) for x in tour]
if best_fitness and fitness <= best_fitness:
logging.debug("No fitness improvement: {0}. Exit EC.".\
format(best_fitness))
break
best_tour, best_fitness = tour, fitness
logging.debug("Current best fitness: {0}".format(best_fitness))
tour = self.fix_orientation(tour)
i += 1
tour = best_tour
recode = {0: '?', 1: '+', -1: '-'}
tour = [(x, recode[o]) for x, o in tour]
self.tour = tour
for mlg in self.lgs:
mapname, lg = mlg.rsplit("-", 1)
if mapname == pivot:
self.object = "chr{0}".format(lg)
break
def layout(args):
"""
%prog layout query.subject.simple query.seqids subject.seqids
Compute optimal seqids order in a second genome, based on seqids on one
genome, given the pairwise blocks in .simple format.
"""
from jcvi.algorithms.ec import GA_setup, GA_run
p = OptionParser(layout.__doc__)
p.set_beds()
p.set_cpus(cpus=32)
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(not p.print_help())
simplefile, qseqids, sseqids = args
qbed, sbed, qorder, sorder, is_self = check_beds(simplefile, p, opts)
qseqids = qseqids.strip().split(",")
sseqids = sseqids.strip().split(",")
qseqids_ii = dict((s, i) for i, s in enumerate(qseqids))
sseqids_ii = dict((s, i) for i, s in enumerate(sseqids))
blocks = SimpleFile(simplefile).blocks
scores = defaultdict(int)
for a, b, c, d, score, orientation, hl in blocks:
qi, q = qorder[a]
si, s = sorder[c]
qseqid, sseqid = q.seqid, s.seqid
if sseqid not in sseqids:
continue
scores[sseqids_ii[sseqid], qseqid] += score
data = []
for (a, b), score in sorted(scores.items()):
if b not in qseqids_ii:
continue
data.append((qseqids_ii[b], score))
tour = range(len(qseqids))
toolbox = GA_setup(tour)
toolbox.register("evaluate", colinear_evaluate_weights, data=data)
tour, fitness = GA_run(toolbox, ngen=100, npop=100, cpus=opts.cpus)
tour = [qseqids[x] for x in tour]
print ",".join(tour)
def layout(args):
"""
%prog layout query.subject.simple query.seqids subject.seqids
Compute optimal seqids order in a second genome, based on seqids on one
genome, given the pairwise blocks in .simple format.
"""
from jcvi.algorithms.ec import GA_setup, GA_run
p = OptionParser(layout.__doc__)
p.set_beds()
p.set_cpus(cpus=32)
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(not p.print_help())
simplefile, qseqids, sseqids = args
qbed, sbed, qorder, sorder, is_self = check_beds(simplefile, p, opts)
qseqids = qseqids.strip().split(",")
sseqids = sseqids.strip().split(",")
qseqids_ii = dict((s, i) for i, s in enumerate(qseqids))
sseqids_ii = dict((s, i) for i, s in enumerate(sseqids))
blocks = SimpleFile(simplefile).blocks
scores = defaultdict(int)
for a, b, c, d, score, orientation, hl in blocks:
qi, q = qorder[a]
si, s = sorder[c]
qseqid, sseqid = q.seqid, s.seqid
if sseqid not in sseqids:
continue
scores[sseqids_ii[sseqid], qseqid] += score
data = []
for (a, b), score in sorted(scores.items()):
if b not in qseqids_ii:
continue
data.append((qseqids_ii[b], score))
tour = range(len(qseqids))
toolbox = GA_setup(tour)
toolbox.register("evaluate", colinear_evaluate_weights, data=data)
tour, fitness = GA_run(toolbox, ngen=100, npop=100, cpus=opts.cpus)
tour = [qseqids[x] for x in tour]
print ",".join(tour)
def test_ec(input_array, expected):
from jcvi.algorithms.ec import (
GA_setup,
GA_run,
colinear_evaluate,
creator,
make_data,
)
POINTS, SCF = 200, 20
scaffolds = make_data(POINTS, SCF)
toolbox = GA_setup(input_array)
toolbox.register("evaluate", colinear_evaluate, scaffolds=scaffolds)
tour, tour.fitness = GA_run(toolbox, cpus=8)
print(tour, tour.fitness)
assert list(tour) == expected
assert tour.fitness == creator.FitnessMax((200.0, ))
def score(args):
"""
%prog score main_results/ cached_data/ contigsfasta
Score the current LACHESIS CLM.
"""
p = OptionParser(score.__doc__)
p.set_cpus()
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(not p.print_help())
mdir, cdir, contigsfasta = args
orderingfiles = natsorted(iglob(mdir, "*.ordering"))
sizes = Sizes(contigsfasta)
contig_names = list(sizes.iter_names())
contig_ids = dict((name, i) for (i, name) in enumerate(contig_names))
oo = []
# Load contact matrix
glm = op.join(cdir, "all.GLM")
N = len(contig_ids)
M = np.zeros((N, N), dtype=int)
fp = open(glm)
for row in fp:
if row[0] == '#':
continue
x, y, z = row.split()
if x == 'X':
continue
M[int(x), int(y)] = int(z)
fwtour = open("tour", "w")
def callback(tour, gen, oo):
fitness = tour.fitness if hasattr(tour, "fitness") else None
label = "GA-{0}".format(gen)
if fitness:
fitness = "{0}".format(fitness).split(",")[0].replace("(", "")
label += "-" + fitness
print_tour(fwtour, tour, label, contig_names, oo)
return tour
for ofile in orderingfiles:
co = ContigOrdering(ofile)
for x in co:
contig_id = contig_ids[x.contig_name]
oo.append(contig_id)
pf = op.basename(ofile).split(".")[0]
print pf
print oo
tour, tour_sizes, tour_M = prepare_ec(oo, sizes, M)
# Store INIT tour
print_tour(fwtour, tour, "INIT", contig_names, oo)
# Faster Cython version for evaluation
from .chic import score_evaluate_M
callbacki = partial(callback, oo=oo)
toolbox = GA_setup(tour)
toolbox.register("evaluate",
score_evaluate_M,
tour_sizes=tour_sizes,
tour_M=tour_M)
tour, tour.fitness = GA_run(toolbox,
npop=100,
cpus=opts.cpus,
callback=callbacki)
print tour, tour.fitness
break
fwtour.close()
def score(args):
"""
%prog score main_results/ cached_data/ contigsfasta
Score the current LACHESIS CLM.
"""
p = OptionParser(score.__doc__)
p.set_cpus()
opts, args = p.parse_args(args)
if len(args) != 3:
sys.exit(not p.print_help())
mdir, cdir, contigsfasta = args
orderingfiles = natsorted(iglob(mdir, "*.ordering"))
sizes = Sizes(contigsfasta)
contig_names = list(sizes.iter_names())
contig_ids = dict((name, i) for (i, name) in enumerate(contig_names))
oo = []
# Load contact matrix
glm = op.join(cdir, "all.GLM")
N = len(contig_ids)
M = np.zeros((N, N), dtype=int)
fp = open(glm)
for row in fp:
if row[0] == '#':
continue
x, y, z = row.split()
if x == 'X':
continue
M[int(x), int(y)] = int(z)
fwtour = open("tour", "w")
def callback(tour, gen, oo):
fitness = tour.fitness if hasattr(tour, "fitness") else None
label = "GA-{0}".format(gen)
if fitness:
fitness = "{0}".format(fitness).split(",")[0].replace("(", "")
label += "-" + fitness
print_tour(fwtour, tour, label, contig_names, oo)
return tour
for ofile in orderingfiles:
co = ContigOrdering(ofile)
for x in co:
contig_id = contig_ids[x.contig_name]
oo.append(contig_id)
pf = op.basename(ofile).split(".")[0]
print pf
print oo
tour, tour_sizes, tour_M = prepare_ec(oo, sizes, M)
# Store INIT tour
print_tour(fwtour, tour, "INIT", contig_names, oo)
# Faster Cython version for evaluation
from .chic import score_evaluate_M
callbacki = partial(callback, oo=oo)
toolbox = GA_setup(tour)
toolbox.register("evaluate", score_evaluate_M,
tour_sizes=tour_sizes, tour_M=tour_M)
tour, tour.fitness = GA_run(toolbox, npop=100, cpus=opts.cpus,
callback=callbacki)
print tour, tour.fitness
break
fwtour.close()
