def configure_outputs(self):
print('Configuring outputs. output_dir: {}'.format(self.output_dir))
print('log_write: {} dcd_write: {}'.format(self.log_write,
self.dcd_write))
hoomd.meta.dump_metadata(filename=os.path.join(self.output_dir,
'metadata.json'),
indent=2)
deprecated.dump.xml(group=hoomd.group.all(),
filename=os.path.join(self.output_dir,
'start.hoomdxml'),
all=True)
quantities = self.get_log_quantities()
if self.dybond_updater is not None:
quantities.append("bond_percent(A-B)")
#quantities.append("avg_num_failed_bonds")
#quantities.append("avg_num_neighbors")
print('quantities being logged:', quantities)
hoomd.analyze.log(filename=os.path.join(self.output_dir, 'out.log'),
quantities=quantities,
period=self.log_write,
header_prefix='#',
overwrite=True)
dump.dcd(filename=os.path.join(self.output_dir, 'traj.dcd'),
period=self.dcd_write,
overwrite=True)
dump.gsd(filename=os.path.join(self.output_dir, 'data.gsd'),
period=self.dcd_write,
group=hoomd.group.all(),
overwrite=True,
static=['attribute'])
msd_groups = self.get_msd_groups()
deprecated.analyze.msd(groups=msd_groups,
period=self.log_write,
overwrite=True,
filename=os.path.join(self.output_dir,
'msd.log'),
header_prefix='#')
self.silent_remove(
os.path.join(self.output_dir, self.bond_rank_hist_file))
def simulate(traj_filename):
with open(traj_filename, "r") as f:
traj = json.load(f)
seed = traj["seed"]
len_chrom = traj["len_chrom"]
Cent = traj["Cent"]
p_ribo = traj["p_ribo"]
R = traj["R"]
micron = traj["micron"]
data_folder = traj["data_folder"]
# Diffusing elements
N_diffu = traj["N_diffu"]
cut_off_inte = traj["cut_off_inte"]
p_inte = traj["p_inte"]
dt = traj["dt"]
p_origins = traj["p_origins"]
# Yeast case
spb = traj["spb"]
nucleole = traj["nucleole"]
telomere = traj["telomere"]
microtubule_length = traj["microtubule_length"] * micron
diameter_nuc = traj["diameter_nuc"] * micron
special_start = traj["special_start"]
Activ_Origins = traj["Activ_Origins"]
visu = traj["visu"]
dump_hic = traj["dump_hic"]
# Scenari
diff_alone = traj["diff_alone"]
diff_bind_when_free = traj["diff_bind_when_free"]
diff_bind_when_on_DNA = traj["diff_bind_when_on_DNA"]
replicate_DNA = traj["replicate_DNA"]
np.random.seed(seed)
hoomd.context.initialize("--mode=cpu")
if diff_alone:
# Check
assert (diff_bind_when_free is False)
assert (diff_bind_when_on_DNA is False)
# End of parameter
##########################################
#########################################
# Define polymer bonding and positions
Np = len(len_chrom)
assert (len(len_chrom) == len(Cent) == len(p_ribo))
if special_start:
Sim = create_init_conf_yeast(len_chrom=len_chrom,
dist_centro=Cent,
p_ribo=p_ribo,
Radius=R,
Mt=microtubule_length)
else:
Sim = []
spbp = 0 if not spb else 1
Total_particle = sum(len_chrom) + N_diffu * 2 + spbp
list_nuc = [
list(range(start, start + size)) if size != 0 else []
for start, size in p_ribo
]
# print(list_nuc)
# exit()
snapshot = data.make_snapshot(N=Total_particle,
box=data.boxdim(L=2 * R),
bond_types=['polymer'])
spbb = Np if spb else 0
if visu:
spbb = 0
bond_diffu = 0
if diff_bind_when_free:
bond_diffu = N_diffu
snapshot.bonds.resize(sum(len_chrom) - len(len_chrom) + bond_diffu + spbb)
bond_list = ['Mono_Mono', 'Diff_Diff', 'Mono_Diff']
if spb:
bond_list += ["Spb_Cen"]
if nucleole:
bond_list += ["Mono_Nuc", "Nuc_Nuc"]
snapshot.bonds.types = bond_list
plist = ['Mono', 'Ori', 'Diff', 'A_Ori', 'P_Ori', 'S_Diff', 'F_Diff']
if spb:
plist.append("Spb")
if nucleole:
plist += ['Nuc', 'A_Nuc', 'P_Nuc']
if telomere:
plist += ["Telo"]
snapshot.particles.types = plist
offset_bond = 0
offset_particle = 0
lPolymers = []
################################################
# Polymer chains
Cen_pos = []
for i in range(Np):
found_cen = False
npp = len_chrom[i] # Number of particles
# Position of origin of replication
pos_origins = p_origins[i]
if Sim == []:
initp = 2 * np.random.rand(3) - 1
else:
# print(i)
initp = Sim.molecules[i].coords[0]
for p in range(npp - 1):
inuc = 0
if nucleole:
if p in list_nuc[i]:
inuc += 1
if p + 1 in list_nuc[i]:
inuc += 1
snapshot.bonds.group[offset_bond + p] = [
offset_particle + p, offset_particle + p + 1
]
if inuc == 0:
snapshot.bonds.typeid[offset_bond + p] = bond_list.index(
'Mono_Mono') # polymer_A
if inuc == 1:
snapshot.bonds.typeid[offset_bond + p] = bond_list.index(
'Mono_Nuc') # polymer_A
if inuc == 2:
snapshot.bonds.typeid[offset_bond + p] = bond_list.index(
'Nuc_Nuc') # polymer_A
offset_bond += npp - 1
for p in range(npp):
# print(offset_bond, offset_bond + p)
if Sim == []:
new = 2 * (2 * np.random.rand(3) - 1)
while linalg.norm(initp + new) > R - 1:
new = 2 * (2 * np.random.rand(3) - 1)
initp += new
else:
initp = Sim.molecules[i].coords[p]
snapshot.particles.position[offset_particle + p] = initp
if p in pos_origins:
snapshot.particles.typeid[offset_particle + p] = plist.index(
'Ori') # Ori
else:
snapshot.particles.typeid[offset_particle + p] = plist.index(
'Mono') # A
if spb and p == Cent[i]:
Cen_pos.append(offset_particle + p)
found_cen = True
if nucleole and p in list_nuc[i]:
snapshot.particles.typeid[offset_particle +
p] = plist.index('Nuc')
if telomere and (p == 0 or p == npp - 1):
snapshot.particles.typeid[offset_particle +
p] = plist.index('Telo')
lPolymers.append(
Polymer(i, offset_particle, offset_particle + npp - 1,
[po + offset_particle for po in pos_origins]))
offset_particle += npp
assert (found_cen == spb)
phic = 0
if dump_hic:
phic = 0 + offset_particle - 1
###################################################
# SPD
if spb:
tag_spb = 0 + offset_particle
# print(tag_spb)
# print(snapshot.particles[offset_particle])
snapshot.particles.position[offset_particle] = [-R + 0.1, 0, 0]
snapshot.particles.typeid[offset_particle] = plist.index('Spb')
offset_particle += 1
if not visu:
for i in range(Np):
# print(offset_particle - 1, Cen_pos[i])
snapshot.bonds.group[offset_bond] = [
offset_particle - 1, Cen_pos[i]
]
snapshot.bonds.typeid[offset_bond] = bond_list.index(
'Spb_Cen') # polymer_A
offset_bond += 1
############################################################
# Diffusing elements
# Defining useful classes
# Defining particles and bonds for the simulation
for i in range(N_diffu):
npp = 2 # Number of particles
initp = (R - 2) * (2 * np.random.rand(3) - 1)
while linalg.norm(initp) > R - 1:
initp = (R - 2) * (2 * np.random.rand(3) - 1)
if diff_bind_when_free:
for p in range(npp - 1):
snapshot.bonds.group[offset_bond + p] = [
offset_particle + p, offset_particle + p + 1
]
snapshot.bonds.typeid[offset_bond + p] = bond_list.index(
'Diff_Diff') # Diff_Diff
offset_bond += npp - 1
for p in range(npp):
# print(offset_bond, offset_bond + p)
if diff_bind_when_free:
new = 2 * (2 * np.random.rand(3) - 1)
while linalg.norm(initp + new) > R - 1:
new = 2 * (2 * np.random.rand(3) - 1)
# print(initp,new,R,linalg.norm(initp + new))
# exit()
initp += new
else:
initp = (R - 1) * (2 * np.random.rand(3) - 1)
snapshot.particles.position[offset_particle + p] = initp
snapshot.particles.typeid[offset_particle + p] = plist.index(
"Diff") # Diffu
offset_particle += npp
# Load the configuration
for i, p in enumerate(snapshot.bonds.group):
if p[0] == p[1]:
print(i, p)
system = init.read_snapshot(snapshot)
for i, p in enumerate(system.particles):
# print(p)
# exit()
assert p.tag == i
for i, b in enumerate(system.bonds):
if b.a == b.b:
print(b.a, b.b)
raise
# print(p)
# exit()
assert b.tag == i
###############################################
###############################################
# Defining force field:
harmonic = md.bond.harmonic()
harmonic.bond_coeff.set(bond_list, k=330.0, r0=1)
harmonic.bond_coeff.set('Mono_Diff', k=10.0, r0=1)
if spb:
harmonic.bond_coeff.set('Spb_Cen', k=1000.0, r0=microtubule_length)
if nucleole:
harmonic.bond_coeff.set('Nuc_Nuc', k=330, r0=diameter_nuc)
harmonic.bond_coeff.set('Mono_Nuc',
k=330,
r0=diameter_nuc / 2. + 1. / 2)
nl = md.nlist.tree(r_buff=0.4, check_period=1)
# Potential for warmup
gauss = md.pair.gauss(r_cut=3.0, nlist=nl)
gauss.pair_coeff.set(plist, plist, epsilon=1.0, sigma=1.0)
if nucleole:
for ip1, p1 in enumerate(plist):
for p2 in plist[ip1:]:
inuc = 0
if "Nuc" in p1:
inuc += 1
if "Nuc" in p2:
inuc += 1
if inuc == 1:
gauss.pair_coeff.set(p1,
p2,
epsilon=.5,
sigma=0.5 + diameter_nuc / 2.,
r_cut=(0.5 + diameter_nuc / 2.) * 3)
if inuc == 2:
gauss.pair_coeff.set(p1,
p2,
epsilon=1.0,
sigma=diameter_nuc,
r_cut=3 * diameter_nuc)
# gauss.pair_coeff.set('A', 'A', epsilon=1.0, sigma=1.0)
# gauss.pair_coeff.set('A', 'A', epsilon=1.0, sigma=1.0)
# Spherical confinement
sphere = md.wall.group()
sphere.add_sphere(r=R, origin=(0.0, 0.0, 0.0), inside=True)
wall_force_slj = md.wall.slj(sphere, r_cut=3.0)
wall_force_slj.force_coeff.set(plist, epsilon=1.0, sigma=1.0, r_cut=1.12)
if nucleole:
wall_force_slj.force_coeff.set('Nuc',
epsilon=1.0,
sigma=diameter_nuc,
r_cut=diameter_nuc * 1.12)
if telomere:
wall_force_slj.force_coeff.set(plist, epsilon=2.0, sigma=1.5, r_cut=3)
# Group;
all_beads = group.all()
if spb:
Spb_g = group.tag_list(name="Spb", tags=[tag_spb])
pspb = [p.position for p in Spb_g]
print(pspb)
all_move = group.difference(name="move", a=all_beads, b=Spb_g)
else:
all_move = all_beads
# Log
logger = analyze.log(filename=data_folder + 'mylog.log',
period=1000,
quantities=[
'temperature', 'potential_energy',
'kinetic_energy', 'volume', 'pressure'
],
overwrite=True)
# Warmup
converged = False
dt = 0.005
while not converged and not visu:
try:
method = md.integrate.mode_minimize_fire(group=all_move, dt=dt)
while not (method.has_converged()):
if spb:
pspb = [p.position for p in Spb_g]
"""
print(pspb)
for cen in Cen_pos:
cent_tmp = system.particles[cen]
print(cent_tmp.position)
print(linalg.norm(np.array(pspb[0])-np.array(cent_tmp.position)))
print(R * microtubule_length)
"""
# exit()
hoomd.run(100)
converged = True
except:
converged = False
dt /= 2.
print(dt)
# Restore positions
for ip, p in enumerate(snapshot.particles.position):
system.particles[ip].position = p
"""
gauss.disable()
slj=md.pair.slj(r_cut=2, nlist=nl)
slj.pair_coeff.set(plist,plist,sigma=1,epsilon=1,r_cut=1.12)
print("Second minimizing")
method=md.integrate.mode_minimize_fire(group=all_beads,dt=0.05)
while not(method.has_converged()):
hoomd.run(100)
"""
# hoomd.run(1000000)
# method.disable()
# Dumping
if visu:
xml = deprecated.dump.xml(filename=data_folder + "atoms.hoomdxml",
period=None,
group=all_beads,
vis=True)
exit()
# gsd = dump.gsd(filename=data_folder + "atoms.gsd",period=None,group=all_beads)
dcd = dump.dcd(filename=data_folder + 'poly.dcd',
period=100,
overwrite=True)
# Dynamics
t0 = time.time()
md.integrate.mode_standard(dt=0.01)
method = md.integrate.langevin(group=all_move, kT=1, seed=seed)
snp = system # .take_snapshot()
def Change_type(typep, particle_list, snp, msg=""):
# print(particle_list)
for p in particle_list:
snp.particles[p].type = typep
if particle_list != [] and msg != "":
print(msg)
def Bind(typeb, bondlist, snp):
btags = []
for b1, b2 in bondlist:
btags.append(snp.bonds.add(typeb, b1, b2))
return btags
def Release(btags, snp):
for bt in btags:
snp.bonds.remove(bt)
def AddParticle(position, type):
snp.particles.add(type)
snp.particles[-1].position = position
def Shift(bonds, snp):
for tag, new in bonds:
b = snp.bonds.get(tag)
btype = "" + b.type
fork = b.b + 0
snp.bonds.remove(tag)
# print(b.type)
snp.bonds.add(btype, new, fork)
# print(new,b)
# print(dir(snp.bonds))
# b.a = new
group_diffu = group.type(name="Diff", type='Diff')
if Activ_Origins != []:
group_origin = group.type(name="Activ_Ori", type=Activ_Origins[0])
if len(Activ_Origins) > 1:
for t in Activ_Origins[1:]:
group_origin = group.union(name="Activ_origin",
a=group_origin,
b=group.type(name="tmp", type=t))
r_hic = []
if dump_hic:
group_hic = group.tags(name="hic", tag_min=0, tag_max=phic)
# nl.tune(warmup=1,steps=1000)
for i in range(100):
# Chek that the microtubule length is correct
if spb:
for cen in Cen_pos:
cent_tmp = system.particles[cen]
# print(cent_tmp.position)
d = linalg.norm(
np.array(pspb[0]) - np.array(cent_tmp.position))
if d > 2 * microtubule_length:
print("MT too long", d)
exit()
# Dump the Hi-Cs
# system.restore_snapshot(snp)
hoomd.run(1000)
if dump_hic:
ph = np.array([p.position for p in group_hic])
D = cdist(ph, ph)
D[D < 2] = 1
D[D >= 2] = 0
np.fill_diagonal(D, 0)
if r_hic != []:
r_hic += D
else:
r_hic = D
np.save(data_folder + "/hic", r_hic)
# snp = system.take_snapshot()
# update the position of the monomer by updating bonds
for iP, P in enumerate(lPolymers):
verbose = False
# if iP == 9:
# verbose = True
bind_diff, diff_diff, shifted_bonds, \
passivated_origin, to_release, alone = P.increment_time(
1, verbose)
Change_type('P_Ori', passivated_origin, snp,
msg="") # Passivated origin
if not diff_alone:
Shift(shifted_bonds, snp)
# Bond tags to release (Alone particle)
Release(to_release, snp)
if diff_bind_when_free:
# Pair of diffu to attach
Bind("Diff_Diff", bind_diff, snp)
# We cannot use the single diff anymore
Change_type("S_Diff", alone, snp)
# Change type for pair of diff diff
Change_type("Diff", diff_diff, snp)
group_diffu.force_update()
group_origin.force_update()
# Update Type because of (Ori to passivated)
# Update group
# Find new interacting particles
# First check if Dimer are close from one origin
p_diffu = np.array([p.position for p in group_diffu])
tag_diffu = [p.tag for p in group_diffu]
p_origin = np.array([p.position for p in group_origin])
tag_origin = [p.tag for p in group_origin]
if tag_diffu != [] and tag_origin != []:
distances = cdist(p_diffu, p_origin)
print(distances.shape)
# Reorder the distances with the dimer tags
Indexes = []
PTags = []
# t0 = time.time()
Btags = []
# Groups Diff-Diff by bond to compute the distances
if diff_bind_when_free:
for b in system.bonds:
if b.type == 'Diff_Diff' and system.particles[
b.a].type == 'Diff':
Indexes.append(tag_diffu.index(b.a))
Indexes.append(tag_diffu.index(b.b))
Btags.append(b.tag)
PTags.append([b.a, b.b])
# print(time.time() -t0)
d2 = distances[Indexes][::2] / 2 + distances[Indexes][1::2] / 2
else:
n_diffu = len(tag_diffu)
Indexes = list(range(n_diffu))
Btags = [None] * n_diffu
PTags = [[t] for t in tag_diffu]
d2 = distances[Indexes]
activated = []
for iD, (btag, ptags) in enumerate(zip(Btags, PTags)):
# print(d2.shape)
# print(d2[iD])
for iorigin, di in enumerate(d2[iD]):
if iorigin in activated:
# Needed because we don't want an origin to be activated
# twice
continue
if di > cut_off_inte:
continue
if np.random.rand() > p_inte:
continue
for P in lPolymers:
if not P.has_origin(tag_origin[iorigin]):
continue
if diff_bind_when_free and \
not diff_bind_when_on_DNA:
Release([btag], snp) # Break the dimer
btag = None # We need btag only in the case where they stays attached
if not diff_alone:
# Or attached separatly or already bound:
if diff_bind_when_free:
# We are sure they are two and we can
# start
Change_type('F_Diff', ptags,
snp) # Diffusive element attached
particular_origin = tag_origin[iorigin]
new_btags = Bind(
"Mono_Diff",
[[particular_origin, ptags[0]],
[particular_origin, ptags[1]]], snp)
Change_type('A_Ori', [particular_origin], snp)
activated.append(iorigin)
P.add_fork(ptags, particular_origin, new_btags,
btag)
else:
Change_type('F_Diff', ptags,
snp) # Diffusive element attached
particular_origin = tag_origin[iorigin]
new_btags = Bind(
"Mono_Diff",
[[particular_origin, ptags[0]]], snp)
start = P.attach_one_diff(
ptags[0], particular_origin, new_btags[0])
if start:
# get particles involves
p1, p2 = P.get_diff_at_origin(
particular_origin)
if diff_bind_when_on_DNA:
btag = Bind("Diff_Diff",
[[p1[0], p2[0]]], snp)[0]
Change_type('A_Ori', [particular_origin],
snp)
P.add_fork([p1[0], p2[0]],
particular_origin,
[p1[1], p2[1]], btag)
else:
# start when touched and release
particular_origin = tag_origin[iorigin]
activated.append(iorigin)
Change_type('A_Ori', [particular_origin], snp)
P.add_fork([None, None], particular_origin,
[None, None], None)
break
# If we arrive there it means that one interaction has beeen
# found
break
# t0 = time.time()
with open(data_folder + "polymer_timing.dat", "wb") as f:
cPickle.dump(lPolymers, f)
# print(time.time() -t0)
# Then if it is the case attach them according to p law to the origin
print(gauss.get_energy(all_beads), wall_force_slj.get_energy(all_beads))
print(time.time() - t0)
def simulate(traj):
seed = traj["seed"]
micron = traj["micron"]
data_folder = traj["data_folder"]
# Diffusing elements
cut_off_inte = traj["cut_off_inte"]
p_inte = traj["p_inte"]
p_off = traj["p_off"]
sim_dt = traj["sim_dt"]
fork_speed = traj["fork_speed"]
dt_speed = traj["dt_speed"]
dscale = traj["dscale"]
# Yeast case
spb = traj["spb"]
assert (type(spb) == bool)
microtubule_length = traj["microtubule_length"] * micron
Activ_Origins = traj["Activ_Origins"]
visu = traj["visu"]
dump_hic = traj["dump_hic"]
two_types = traj.get("two_types", False)
r_diffu = traj.get("diameter_diffu", 1) / 2
# Scenari
diff_alone = traj["diff_alone"]
diff_bind_when_free = traj["diff_bind_when_free"]
diff_bind_when_on_DNA = traj["diff_bind_when_on_DNA"]
assert (type(diff_alone) == bool)
assert (type(diff_bind_when_on_DNA) == bool)
assert (type(diff_bind_when_free) == bool)
# Simulation parameters
n_steps = traj["n_steps"]
length_steps = traj["length_steps"]
benchmark = traj["benchmark"]
warmup = traj["warmup"]
dcd_period = traj["dcd_period"]
dump_inte = traj.get("dump_inte", False)
np.random.seed(seed)
hoomd.context.initialize() # "--mode=cpu ")
ramp_type = traj.get("ramp_type", "exp")
ramp = traj.get("ramp_time", 3)
if diff_alone:
# Check
assert (diff_bind_when_free is False)
assert (diff_bind_when_on_DNA is False)
# End of parameter
##########################################
#########################################
# Define polymer bonding and positions
snapshot, phic, tag_spb, bond_list, plist, Cen_pos, lPolymers, list_ori, p_tag_list = \
create_initial_configuration(traj)
system = init.read_snapshot(snapshot)
for i, p in enumerate(system.particles):
# print(p)
# exit()
assert p.tag == i
for i, b in enumerate(system.bonds):
if b.a == b.b:
print(b.a, b.b)
raise
assert b.tag == i
###############################################
###############################################
# Defining force field:
all_beads, all_move, Spb_g, nl = force_field(traj,
bond_list=bond_list,
plist=plist,
tag_spb=tag_spb,
two_types=two_types)
# Log
if not visu:
logger = analyze.log(filename=data_folder + 'mylog.log',
period=1000,
quantities=[
'temperature', 'potential_energy',
'bond_harmonic_energy',
'external_wall_lj_energy',
"pair_table_energy", 'kinetic_energy',
'volume', 'pressure'
],
overwrite=True)
# Warmup
minimize(traj, all_move, system, snapshot, Spb_g, Cen_pos,
microtubule_length)
# Dumping
if visu:
xml = deprecated.dump.xml(filename=data_folder + "atoms.hoomdxml",
period=None,
group=all_beads,
vis=True)
# xml.disable()
return
# gsd = dump.gsd(filename=data_folder + "atoms.gsd",period=None,group=all_beads)
# Dynamics
def Change_type(typep, particle_list, snp, msg=""):
# print(particle_list)
for p in particle_list:
if "Ori" in typep:
# Remove it from the list activated
# list_ori.remove(p)
pass
else:
snp.particles[p].type = typep
if particle_list != [] and msg != "":
print(msg)
def Bind(typeb, bondlist, snp):
btags = []
for b1, b2 in bondlist:
btags.append(snp.bonds.add(typeb, b1, b2))
return btags
def Release(btags, snp):
for bt in btags:
snp.bonds.remove(bt)
def AddParticle(position, type):
snp.particles.add(type)
snp.particles[-1].position = position
def Shift(bonds, snp):
for tag, new in bonds:
b = snp.bonds.get(tag)
btype = "" + b.type
fork = b.b + 0
snp.bonds.remove(tag)
# print(b.type)
snp.bonds.add(btype, new, fork)
# print(new,b)
# print(dir(snp.bonds))
# b.a = new
snp = system # .take_snapshot()
if ramp_type == "exp":
for couple in p_tag_list:
Change_type("I_Diff", couple, snp)
group_diffu = group.type(name="Diff", type='Diff')
# nl.tune(warmup=1,steps=1000)
# Small warmup
t0 = time.time()
md.integrate.mode_standard(dt=sim_dt)
method = md.integrate.langevin(group=all_move,
kT=1,
seed=seed,
dscale=False)
print(plist)
for p in ['Diff', 'S_Diff', 'F_Diff', "I_Diff"]:
print(p, dscale * r_diffu)
method.set_gamma(p, dscale * r_diffu)
for p in ['Mono', 'Ori']:
method.set_gamma(p, dscale)
if two_types:
method.set_gamma(p, dscale)
# exit()
if benchmark:
print(nl.tune(warmup=4000, r_min=0.3, r_max=0.8, jumps=5, steps=5000))
return
"""
md.integrate.mode_standard(dt=sim_dt / 4)
hoomd.run(100)
md.integrate.mode_standard(dt=sim_dt / 2)
hoomd.run(100)
md.integrate.mode_standard(dt=sim_dt)
"""
if warmup != 0:
hoomd.run(warmup)
dcd = dump.dcd(filename=data_folder + 'poly.dcd',
period=dcd_period,
overwrite=True)
r_hic = []
# if dump_hic:
group_hic = group.tags(name="hic", tag_min=0, tag_max=phic)
if dump_inte:
r_inte = []
Ndiff_libre_t = []
N_diffu = traj["N_diffu"]
offset_diff = np.min(p_tag_list)
print(offset_diff)
record_diffusing = [Diffusing(d) for d in np.arange(N_diffu * 2)]
global timeit
timeit = True
global t0
t0 = time.time()
def Timeit(where=""):
global timeit
global t0
if timeit:
if where == "":
print(time.time() - t0)
else:
print(where, time.time() - t0)
t0 = time.time()
previous_actifs = 0
for i in range(n_steps):
# Chek that the microtubule length is correct
if spb:
for cen in Cen_pos:
cent_tmp = system.particles[cen]
# print(cent_tmp.position)
pspb = [p.position for p in Spb_g]
d = linalg.norm(
np.array(pspb[0]) - np.array(cent_tmp.position))
if d > 2 * microtubule_length:
print("MT too long", d)
exit()
# Dump the Hi-Cs
Timeit()
# system.restore_snapshot(snp)
N_actifs = int(len(p_tag_list) * (1 - np.exp(-i * dt_speed / ramp)))
print(previous_actifs, N_actifs)
for couple in p_tag_list[previous_actifs:N_actifs]:
Change_type("Diff", couple, snp)
print("Activated", couple)
previous_actifs = N_actifs
hoomd.run(length_steps // 2, profile=False)
Timeit("After first half")
if dump_hic and i % traj.get("hic_period", 1) == 0:
ph = np.array([p.position for p in group_hic])
D = cdist(ph, ph)
D[D < 2] = 1
D[D >= 2] = 0
np.fill_diagonal(D, 0)
if r_hic != []:
r_hic += D
else:
r_hic = D
if i % 32 == 0:
np.save(data_folder + "/hic", r_hic)
# snp = system.take_snapshot()
# update the position of the monomer by updating bonds
ended = 0
for iP, P in enumerate(lPolymers):
verbose = False
# if iP == 9:
# verbose = True
bind_diff, diff_diff, shifted_bonds, \
passivated_origin, to_release, alone = P.increment_time(
dt_speed, verbose)
###################################################################
# Only to keep track of the diffusing elements
for diff1, diff2 in bind_diff:
record_diffusing[diff1 - offset_diff].end_replication(
i * dt_speed, pos=snp.particles[diff1].position)
record_diffusing[diff2 - offset_diff].end_replication(
i * dt_speed, pos=snp.particles[diff2].position)
for diff in alone:
if record_diffusing[diff - offset_diff].bound:
record_diffusing[diff - offset_diff].end_bound(
i * dt_speed, pos=snp.particles[diff].position)
elif record_diffusing[diff - offset_diff].replicating:
record_diffusing[diff - offset_diff].end_replication(
i * dt_speed, pos=snp.particles[diff].position)
else:
print(diff, record_diffusing[diff - offset_diff].free)
raise
###################################################################
###################################################################
# take care of the bondings
for ori in passivated_origin:
list_ori.remove(ori)
if not diff_alone:
Shift(shifted_bonds, snp)
# Bond tags to release (Alone particle)
Release(to_release, snp)
if diff_bind_when_free:
# Pair of diffu to attach
Bind("Diff_Diff", bind_diff, snp)
# We cannot use the single diff anymore
Change_type("S_Diff", alone, snp)
# Change type for pair of diff diff
Change_type("Diff", diff_diff, snp)
# check for releasing alone binded elements
for ori_not_started in P.get_free_origins():
diff = P.get_diff_at_origin(ori_not_started)
if diff != []:
if np.random.rand() > p_off:
continue
ptag, bond_tag = diff[0]
P.dettach_one_diff(ptag, ori_not_started)
Release([bond_tag], snp)
Change_type("Diff", [ptag], snp)
if P.modules == []:
ended += 1
Timeit("AFter update")
hoomd.run(length_steps // 2, profile=True)
Timeit("AFter second half")
group_diffu.force_update()
p_diffu = np.array([p.position for p in group_diffu])
tag_diffu = [p.tag for p in group_diffu]
p_origin = np.array([snp.particles[ori].position for ori in list_ori])
Ndiff_libre_t.append(len(tag_diffu))
if not (tag_diffu != [] and list_ori != []):
print("No interactions")
# Generate the measures we are interested in
# Matrice interaction DNA / particules
if dump_inte and i % traj.get("inte_period",
1) == 0 and len(group_diffu) != 0:
ph = np.array([p.position for p in group_hic])
pi = np.array([p.position for p in group_diffu])
print(ph.shape, pi.shape)
D = cdist(ph, pi)
D[D < 2] = 0.5 + r_diffu
D[D >= 2] = 0
# np.fill_diagonal(D, 0)
if r_inte != []:
r_inte += D
else:
r_inte = D
if i % 32 == 0:
np.save(data_folder + "/inte", r_inte)
start_replication = False
if tag_diffu != [] and list_ori != [] and start_replication:
distances = cdist(p_diffu, p_origin)
print(distances.shape)
# Reorder the distances with the dimer tags
Indexes = []
PTags = []
# t0 = time.time()
Btags = []
# Groups Diff-Diff by bond to compute the distances
if diff_bind_when_free:
for b in system.bonds:
if b.type == 'Diff_Diff' and system.particles[
b.a].type == 'Diff':
if b.a in tag_diffu:
Indexes.append(tag_diffu.index(b.a))
Indexes.append(tag_diffu.index(b.b))
Btags.append(b.tag)
PTags.append([b.a, b.b])
# print(time.time() -t0)
d2 = distances[Indexes][::2] / 2 + distances[Indexes][1::2] / 2
else:
n_diffu = len(tag_diffu)
Indexes = list(range(n_diffu))
Btags = [None] * n_diffu
PTags = [[t] for t in tag_diffu]
d2 = distances[Indexes]
activated = []
for iD, (btag, ptags) in enumerate(zip(Btags, PTags)):
# print(d2.shape)
# print(d2[iD])
lo = list(range(len(d2[iD])))
np.random.shuffle(lo)
for iorigin in lo:
di = d2[iD][iorigin]
if iorigin in activated:
# Needed because we don't want an origin to be activated
# twice
continue
if di > cut_off_inte:
continue
for P in lPolymers:
if not P.has_origin(list_ori[iorigin]):
continue
if np.random.rand(
) > p_inte * P.o_strength[list_ori[iorigin]]:
continue
if diff_bind_when_free and \
not diff_bind_when_on_DNA:
Release([btag], snp) # Break the dimer
btag = None # We need btag only in the case where they stays attached
if not diff_alone:
# Or attached separatly or already bound:
if diff_bind_when_free:
# We are sure they are two and we can
# start
Change_type('F_Diff', ptags,
snp) # Diffusive element attached
particular_origin = list_ori[iorigin]
new_btags = Bind(
"Mono_Diff",
[[particular_origin, ptags[0]],
[particular_origin, ptags[1]]], snp)
activated.append(0 + iorigin)
P.add_fork(ptags,
particular_origin,
new_btags,
btag,
fork_speed=fork_speed)
for diff in ptags:
record_diffusing[
diff - offset_diff].start_replication(
particular_origin,
i * dt_speed,
pos=snp.particles[diff].position)
else:
Change_type('F_Diff', ptags,
snp) # Diffusive element attached
particular_origin = list_ori[iorigin]
new_btags = Bind(
"Mono_Diff",
[[particular_origin, ptags[0]]], snp)
start = P.attach_one_diff(
ptags[0], particular_origin, new_btags[0])
if start:
# get particles involves
p1, p2 = P.get_diff_at_origin(
particular_origin)
if diff_bind_when_on_DNA:
btag = Bind("Diff_Diff",
[[p1[0], p2[0]]], snp)[0]
activated.append(0 + iorigin)
P.add_fork([p1[0], p2[0]],
particular_origin,
[p1[1], p2[1]],
btag,
fork_speed=fork_speed)
record_diffusing[
p1[0] - offset_diff].start_replication(
particular_origin,
i * dt_speed,
pos=snp.particles[p1[0]].position)
record_diffusing[
p2[0] - offset_diff].start_replication(
particular_origin,
i * dt_speed,
pos=snp.particles[p2[0]].position)
else:
record_diffusing[
ptags[0] - offset_diff].start_bound(
particular_origin,
i * dt_speed,
pos=snp.particles[
ptags[0]].position)
else:
# start when touched and release
particular_origin = list_ori[iorigin]
activated.append(iorigin)
P.add_fork([None, None],
particular_origin, [None, None],
None,
fork_speed=fork_speed)
break
# If we arrive there it means that one interaction has beeen
# found
break
activated.sort()
print(activated)
print(list_ori)
for io in activated[::-1]:
print(io)
list_ori.pop(io)
Timeit("After binding")
# t0 = time.time()
with open(data_folder + "polymer_timing.dat", "wb") as f:
cPickle.dump(lPolymers, f, protocol=2)
with open(data_folder + "Ndiff_libre_t.dat", "wb") as f:
cPickle.dump(Ndiff_libre_t, f, protocol=2)
with open(data_folder + "record_diffusing.dat", "wb") as f:
cPickle.dump(record_diffusing, f, protocol=2)
Timeit("After writing")
if traj.get("early_stop",
True) and list_ori == [] and ended == len(lPolymers):
break
# print(time.time() -t0)
# Then if it is the case attach them according to p law to the origin
# print(gauss.get_energy(all_beads), wall_force_slj.get_energy(all_beads))
print(time.time() - t0)
logger.disable()
method.disable()
dcd.disable()
def minimize(traj,
all_move,
system,
snapshot,
Spb_g,
Cen_pos,
microtubule_length,
maxi=20):
R = traj["R"]
data_folder = traj["data_folder"]
# Diffusing elements
# Yeast case
spb = traj["spb"]
visu = traj["visu"]
# Scenari
converged = False
dt = 0.002
dcd = dump.dcd(filename=data_folder + 'init.dcd',
period=10,
overwrite=True)
it = 0
while not converged and not visu:
try:
if '2.2' in hoomd.__version__:
method = md.integrate.mode_minimize_fire(dt=dt)
nve = md.integrate.nve(group=all_move)
else:
method = md.integrate.mode_minimize_fire(group=all_move, dt=dt)
while not (method.has_converged()):
if spb:
pspb = [p.position for p in Spb_g]
for cen in Cen_pos:
cent_tmp = system.particles[cen]
d = linalg.norm(
np.array(pspb[0]) - np.array(cent_tmp.position))
if d > 2 * microtubule_length:
print("MT too long", d)
raise
"""
print(pspb)
for cen in Cen_pos:
cent_tmp = system.particles[cen]
print(cent_tmp.position)
print(linalg.norm(np.array(pspb[0])-np.array(cent_tmp.position)))
print(R * microtubule_length)
"""
# exit()
hoomd.run(100)
# Check all particles are in the sphere:
for p in system.particles:
if linalg.norm(p.position) > R:
raise
it += 1
if it > maxi:
break
converged = True
if '2.2' in hoomd.__version__:
nve.disable()
except:
if '2.2' in hoomd.__version__:
nve.disable()
converged = False
dt /= 2.
print("Reducing time step", dt)
# Restore positions
for ip, p in enumerate(snapshot.particles.position):
system.particles[ip].position = p
dcd.disable()
#!/usr/bin/env python # Copyright (c) 2020 The Regents of the University of Michigan # All rights reserved. # This software is licensed under the BSD 3-Clause License. from hoomd import deprecated, run, dump, context, group context.initialize() system = deprecated.init.create_random(N=10, phi_p=0.05) deprecated.dump.xml(filename='hoomd.xml', vis=True, group=group.all()) dump.gsd(filename='hoomd.gsd', period=1, overwrite=True, group=group.all()) dump.dcd(filename='dump.dcd', period=1, overwrite=True) run(10)
wall_force_slj.force_coeff.set(plist,
epsilon=1.0,
sigma=1.0,
r_cut=1.12,
mode="shift")
# hoomd.comm.barrier()
microtubule_length = None
Spb_g = None
all_move = group.all()
minimize(traj, all_move, system, snapshot, Spb_g, Cp, microtubule_length)
sim_dt = traj["sim_dt"]
seed = traj["seed"]
md.integrate.mode_standard(dt=sim_dt)
method = md.integrate.langevin(group=all_move, kT=1, seed=seed)
snp = system # .take_snapshot()
md.integrate.mode_standard(dt=sim_dt / 4)
hoomd.run(100)
md.integrate.mode_standard(dt=sim_dt / 2)
hoomd.run(100)
dcd = dump.dcd(filename=data_folder + 'poly.dcd',
period=dcd_period,
overwrite=True)
hoomd.run(1000000)
dcd.disable()
logger.disable()
def simulate(traj):
seed = traj["seed"]
micron = traj["micron"]
data_folder = traj["data_folder"]
# Diffusing elements
cut_off_inte = traj["cut_off_inte"]
p_inte = traj["p_inte"]
p_off = traj["p_off"]
sim_dt = traj["sim_dt"]
dscale = traj["dscale"]
# Yeast case
spb = traj["spb"]
assert (type(spb) == bool)
microtubule_length = traj["microtubule_length"] * micron
visu = traj["visu"]
dump_hic = traj["dump_hic"]
two_types = traj.get("two_types", False)
# Scenari
diff_alone = traj["diff_alone"]
diff_bind_when_free = traj["diff_bind_when_free"]
diff_bind_when_on_DNA = traj["diff_bind_when_on_DNA"]
assert (type(diff_alone) == bool)
assert (type(diff_bind_when_on_DNA) == bool)
assert (type(diff_bind_when_free) == bool)
# Simulation parameters
n_steps = traj["n_steps"]
length_steps = traj["length_steps"]
benchmark = traj["benchmark"]
warmup = traj["warmup"]
dcd_period = traj["dcd_period"]
np.random.seed(seed)
hoomd.context.initialize() # "--mode=cpu ")
if diff_alone:
# Check
assert (diff_bind_when_free is False)
assert (diff_bind_when_on_DNA is False)
# End of parameter
##########################################
#########################################
# Define polymer bonding and positions
snapshot, phic, tag_spb, bond_list, plist, Cen_pos, lPolymers, list_ori, p_tag_list = \
create_initial_configuration(traj)
system = init.read_snapshot(snapshot)
for i, p in enumerate(system.particles):
# print(p)
# exit()
assert p.tag == i
for i, b in enumerate(system.bonds):
if b.a == b.b:
print(b.a, b.b)
raise
assert b.tag == i
###############################################
###############################################
# Defining force field:
all_beads, all_move, Spb_g, nl = force_field(traj,
bond_list=bond_list,
plist=plist,
tag_spb=tag_spb,
two_types=two_types)
# Log
if not visu:
logger = analyze.log(filename=data_folder + 'mylog.log',
period=1000,
quantities=[
'temperature', 'potential_energy',
'bond_harmonic_energy',
'external_wall_lj_energy',
"pair_table_energy", 'kinetic_energy',
'volume', 'pressure'
],
overwrite=True)
# Warmup
minimize(traj, all_move, system, snapshot, Spb_g, Cen_pos,
microtubule_length)
# Dumping
if visu:
xml = deprecated.dump.xml(filename=data_folder + "atoms.hoomdxml",
period=None,
group=all_beads,
vis=True)
# xml.disable()
return
# gsd = dump.gsd(filename=data_folder + "atoms.gsd",period=None,group=all_beads)
# Dynamics
def Change_type(typep, particle_list, snp, msg=""):
# print(particle_list)
for p in particle_list:
if "Ori" in typep:
# Remove it from the list activated
# list_ori.remove(p)
pass
else:
snp.particles[p].type = typep
if particle_list != [] and msg != "":
print(msg)
def Bind(typeb, bondlist, snp):
btags = []
for b1, b2 in bondlist:
btags.append(snp.bonds.add(typeb, b1, b2))
return btags
def Release(btags, snp):
for bt in btags:
snp.bonds.remove(bt)
def AddParticle(position, type):
snp.particles.add(type)
snp.particles[-1].position = position
def Shift(bonds, snp):
for tag, new in bonds:
b = snp.bonds.get(tag)
btype = "" + b.type
fork = b.b + 0
snp.bonds.remove(tag)
# print(b.type)
snp.bonds.add(btype, new, fork)
# print(new,b)
# print(dir(snp.bonds))
# b.a = new
snp = system # .take_snapshot()
# nl.tune(warmup=1,steps=1000)
# Small warmup
t0 = time.time()
md.integrate.mode_standard(dt=sim_dt)
method = md.integrate.langevin(group=all_move,
kT=1,
seed=seed,
dscale=dscale)
if benchmark:
print(nl.tune(warmup=4000, r_min=0.3, r_max=0.8, jumps=5, steps=5000))
return
md.integrate.mode_standard(dt=sim_dt / 4)
hoomd.run(100)
md.integrate.mode_standard(dt=sim_dt / 2)
hoomd.run(100)
md.integrate.mode_standard(dt=sim_dt)
if warmup != 0:
hoomd.run(warmup)
dcd = dump.dcd(filename=data_folder + 'poly.dcd',
period=dcd_period,
overwrite=True)
r_hic = []
if dump_hic:
group_hic = group.tags(name="hic", tag_min=0, tag_max=phic)
global timeit
timeit = True
global t0
t0 = time.time()
def Timeit(where=""):
global timeit
global t0
if timeit:
if where == "":
print(time.time() - t0)
else:
print(where, time.time() - t0)
t0 = time.time()
bonds = []
for i in range(n_steps):
# Chek that the microtubule length is correct
if spb:
for cen in Cen_pos:
cent_tmp = system.particles[cen]
# print(cent_tmp.position)
pspb = [p.position for p in Spb_g]
d = linalg.norm(
np.array(pspb[0]) - np.array(cent_tmp.position))
if d > 2 * microtubule_length:
print("MT too long", d)
exit()
# Dump the Hi-Cs
Timeit()
# system.restore_snapshot(snp)
hoomd.run(length_steps // 2, profile=False)
Timeit("After first half")
if dump_hic and i % traj.get("hic_period", 1) == 0:
ph = np.array([p.position for p in group_hic])
D = cdist(ph, ph)
D[D < 2] = 1
D[D >= 2] = 0
np.fill_diagonal(D, 0)
if r_hic != []:
r_hic += D
else:
r_hic = D
if i % 32 == 0:
np.save(data_folder + "/hic", r_hic)
# snp = system.take_snapshot()
# Bond tags to release (Alone particle)
remv = []
for ib, b in enumerate(bonds):
if np.random.rand() > p_off:
continue
p_tag_list.append(b[2])
list_ori.append(b[1])
Release([b[0]], snp)
remv.append(ib)
remv.sort()
for ib in remv[::-1]:
bonds.pop(ib)
Timeit("AFter update")
hoomd.run(length_steps // 2, profile=True)
Timeit("AFter second half")
# Update Type because of (Ori to passivated)
# Update group
# Find new interacting particles
# First check if Dimer are close from one origin
print("LAAAAAAAAAAAAAAA", p_tag_list, list_ori, bonds)
p_diffu = np.array(
[snp.particles[diff].position for diff in p_tag_list])
p_origin = np.array([snp.particles[ori].position for ori in list_ori])
if not (p_tag_list != [] and list_ori != []):
print("No interactions")
if p_tag_list != [] and list_ori != []:
distances = cdist(p_diffu, p_origin)
print(distances.shape)
# Reorder the distances with the dimer tags
# Groups Diff-Diff by bond to compute the distances
activated_ori = []
activated_p = []
for iD, ptag in enumerate(p_tag_list):
# print(d2.shape)
# print(d2[iD])
lo = list(range(len(distances[iD])))
np.random.shuffle(lo)
for iorigin in lo:
di = distances[iD][iorigin]
if iorigin in activated_ori:
# Needed because we don't want an origin to be activated
# twice
continue
if di > cut_off_inte:
continue
if np.random.rand() > p_inte:
continue
activated_ori.append(iorigin + 0)
activated_p.append(iD)
new_btags = Bind("Mono_Diff", [[list_ori[iorigin], ptag]],
snp)
bonds.append([new_btags[0], list_ori[iorigin], ptag])
break
activated_ori.sort()
for ori in activated_ori[::-1]:
list_ori.pop(ori)
activated_p.sort()
for p in activated_p[::-1]:
p_tag_list.pop(p)
Timeit("After binding")
# t0 = time.time()
# print(time.time() -t0)
# Then if it is the case attach them according to p law to the origin
# print(gauss.get_energy(all_beads), wall_force_slj.get_energy(all_beads))
print(time.time() - t0)
logger.disable()
method.disable()
dcd.disable()