def main():
with open("../mmodes/ModelsInput/media.json") as jdata:
media = json.load(jdata)[0]
# 1) instantiate Consortium
# if "manifest" contains a non-empty string, it will generate COMETS-like ouput
cons = mmodes.Consortium(stcut=1e-7,
mets_to_plot=["ac[e]", "thr_L[e]"],
v=1,
manifest="COMETS_manifest.txt")
# 2) instantiate dMetabolites
# for instance, https://www.ncbi.nlm.nih.gov/pubmed/18791026?dopt=Abstract
glc = mmodes.dMetabolite(id="glc_D[e]", Km=14.8, Vmax=0.13)
# 3) add model
# model from AGORA. A single strain to agilize example.
cons.add_model(
"../mmodes/ModelsInput/BGN4_eu.xml",
0.0003,
solver="glpk",
method="fba",
dMets={glc.id:
glc}) # Bifidobacterium_adolescentis_ATCC_15703.xml (euavg)
# 4) instantiate media
cons.media = cons.set_media(media, True)
# 5) run it, plotting the output
cons.run(maxT=10,
outp="plot_example.png",
outf="tab_example.tsv",
verbose=True,
integrator="fea",
stepChoiceLevel=(0.008, 0.5, 10000))
# 6) print stuff on screen
for mod in cons.models:
print(mod, cons.models[mod].volume.q, sep=" -> ")
print("Glucose", str(cons.media["glc_D[e]"]), sep=" -> ")
print("Acetate", str(cons.media["ac[e]"]), sep=" -> ")
print()
def runn(md=""):
lock.acquire()
print("Starting simulation on", md)
lock.release()
# make working directory based on argument
subprocess.run(["mkdir", md])
# clean directory if needed
files = [
f'{md}/{f}' for f in os.listdir(".") if
f.find("log_template.txt") != -1 or f[-3:] == "tsv" or f[-3:] == 'png'
]
if files:
subprocess.run(["rm"] + files)
# PARAMETERS
# 1) Common parameters of simulation
media_file = "media.json"
mod_dir = "/home/cleanapp/draft/ModelsInput/"
intervl = 1 # time of simulation between perturbations; total time will be 2h
# 2) random parameters of simulation
# 2.1) Medium
with open(mod_dir + media_file) as json_file:
gen_media = json.load(json_file)
# 2.2) Biomasses
brand = lambda: random.uniform(0.000013, 0.000055)
biomasses = [
brand()
] # BGJ: Biomass for athrobacter that always will be in the consortium
# we want to take all the possibilities with *equal* probabilities
# BGJ: add 2 additional values to the biomasses vector, that could be only one, both ones or any.
chosen = random.random()
if chosen > 0.75:
biomasses += [brand(), 0]
elif chosen > 0.5:
biomasses += [0, brand()]
elif chosen > 0.25:
biomasses += [brand(), brand()]
else:
biomasses += [0, 0]
ar, hb, hl = biomasses # BGJ: split biomasses in 3 different variables
# SIMULATION
# 1) instantiate Consortium
volume_petri = 2 * 3.141593 * 45 * 45 * 15 * 1e-6 # 2pi*r²*h -> mm³ to L
cons = mmodes.Consortium(stcut=1e-8,
v=volume_petri,
comets_output=False,
manifest=f"{md}/fluxes.tsv",
work_based_on="id",
max_growth=10,
mets_to_plot=["cpd03959_e0", "cpd00027_e0"],
title="Atrazine " + md)
# 2.1) add models, with random biomass
cons.add_model(mod_dir + "Arthrobacter_CORRECTED.json",
float(ar),
solver="glpk",
method="fba")
cons.add_model(mod_dir + "Halobacillus_sp_CORRECTED.json",
float(hb),
solver="glpk",
method="fba")
cons.add_model(mod_dir + "Halomonas_stevensii_CORRECTED.json",
float(hl),
solver="glpk",
method="fba")
st_ids = [id for id in cons.models if not id.startswith("k")]
print(f"Models were loaded on {md}.")
# 2.2) define initial medium => medium or exudate, already in JSON file
root = deepcopy(gen_media[1])
del (gen_media[random.randint(0, 1)]) # choose medium or exudate
# 2.3) add several random perturbations : strain | both | none
fix_biomass = 0.000034 # BGJ: fix biomass to add in the perturbation as probiotics
for pert in range(3): # BGJ: 2019.11.08
# perturbations always carry atrazine
gen_media.append({
"MEDIA": {
"cpd03959_e0": 0.15
},
"PERTURBATION": "ATRAZINE"
})
chosen = random.random()
if chosen > 0.8:
gen_media[-1]["MEDIA"][st_ids[0]] = fix_biomass * 100
gen_media[-1]["PERTURBATION"] = st_ids[0]
elif chosen > 0.6: # 2019.11.12: BGJ: to increase biomass 2nd strain: st_ids[1]
gen_media[-1]["MEDIA"][st_ids[1]] = fix_biomass * 100
gen_media[-1]["PERTURBATION"] = st_ids[1]
elif chosen > 0.4:
gen_media[-1]["PERTURBATION"] = st_ids[0] + "_" + st_ids[1]
gen_media[-1]["MEDIA"][st_ids[0]] = fix_biomass * 100
gen_media[-1]["MEDIA"][st_ids[1]] = fix_biomass * 100
# elif chosen > 0.2:
# gen_media[-1]["MEDIA"] = root["MEDIA"]
# gen_media[-1]["PERTURBATION"] = "ROOT_EXUDATE"
else:
gen_media[-1]["MEDIA"]['cpd00009_e0'] = 1
gen_media[-1]["PERTURBATION"] = 'PHOSPHATE'
#else: none perturbation (only atrazine)
# 2.3) add 2nd perturbation (nothing)
# we need this to evaluate the last state of the consortium because the
# reward function in MDPbiome for this particular case evaluates degradation of
# atrazine 1 h after the simulation
gen_media.append({
"MEDIA": {
"cpd03959_e0": 0.15
},
"PERTURBATION": "ATRAZINE"
})
# 3) Write log
lock.acquire()
log(cons, gen_media)
lock.release()
it = 1
t_pers = []
pers = []
for mper in gen_media:
if it == 1:
# 4) instantiate media
cons.media = cons.set_media(mper["MEDIA"], True)
else:
for k in mper["MEDIA"]: # not needed at all...
if mper["MEDIA"][k] == 0:
cons.media[k] = 0
cons.add_mets(mper["MEDIA"], True)
pers.append(mper["PERTURBATION"])
# 5) run it
t_pers.append(cons.T[-1])
cons.run(verbose=False,
plot=False,
maxT=intervl + cons.T[-1],
integrator="FEA",
stepChoiceLevel=(0.00027, 0.5, 100000.),
outp=f'{md}/{md}_plot.png',
outf=f'{md}/plot.tsv')
it += 1
txpers = {t_pers[i]: pers[i] for i in range(len(t_pers))}
# 6. Save simulation
with open(f'{md}/cons.p', 'wb') as f:
pickle.dump(cons, f)
with open(f'{md}/txpers.p', 'wb') as f:
pickle.dump(txpers, f)
tsv_filter(f'{md}/plot.tsv',
f'{md}/fluxes.tsv',
txpers,
inplace=False,
v=cons.v)
if os.path.isfile(f'{md}/fluxes_filtered.tsv'):
os.unlink(f'{md}/fluxes.tsv')
mmodes.vis.plot_comm(cons)
lock.acquire()
print(f"\033[1;32;40mProcess with directory {md} out!\033[0m")
lock.release()
del (cons)
del (txpers)
del (gen_media)
return
def runn(md=""):
lock.acquire()
print("Starting simulation on", md)
lock.release()
# make working directory based on argument
subprocess.run(["mkdir", md])
# clean directory if needed
files = [
f'{md}/{f}' for f in os.listdir(".") if
f.find("log_template.txt") != -1 or f[-3:] == "tsv" or f[-3:] == 'png'
]
if files:
subprocess.run(["rm"] + files)
# PARAMETERS
# 1) Common parameters of simulation
media_file = "4_medios.json"
mod_dir = "/home/javi/Documentos/Root_con_P/" #Directorio donde está el modelo
intervl = 1 # time of simulation between perturbations; total time will be 2h
# 2) random parameters of simulation
# 2.1) Medium
with open(mod_dir + media_file) as json_file:
gen_media = json.load(
json_file
) #Abrimos el contenido del archivo y lo metemos en gen_media
# 2.2) Biomasses
brand = lambda: random.uniform(
0.000013, 0.000055
) #Función lambda, se crea así y hace lo que está detras de ':'
#cada vez que llamemos a lambda se nos generará un número random entre ese intervalo
biomasses = [
brand()
] # BGJ: Biomass for athrobacter that always will be in the consortium
# we want to take all the possibilities with *equal* probabilities
# BGJ: add 2 additional values to the biomasses vector, that could be only one, both ones or any.
#Este bloque lo que hace es que de manera aleatoria genera unas biomasas iniciales distintas para cada microorganismo
#Para que cada experimento sea distinto. Por ello, siempre habrá athrobacter (cantidad aleatoria), pero las cantidades de las otras dos
#Las ponemos de manera aleatoria, poniendo solo uno de ellos, los dos o ninguno.
chosen = random.random() #Generamos un número aleatorio entre cero y uno.
if chosen > 0.75:
biomasses += [brand(), 0]
elif chosen > 0.5:
biomasses += [0, brand()]
elif chosen > 0.25:
biomasses += [brand(), brand()]
else:
biomasses += [0, 0]
#Ponemos cada uno de los valores de biomasa en tres variables que usaremos después.
ar, hb, hl = biomasses # BGJ: split biomasses in 3 different variables
# SIMULATION
# 1) instantiate Consortium
volume_petri = 2 * 3.141593 * 45 * 45 * 15 * 1e-6 # 2pi*r²*h -> mm³ to L
cons = mmodes.Consortium(stcut=1e-8,
v=volume_petri,
comets_output=False,
manifest=f"{md}/fluxes.tsv",
work_based_on="id",
max_growth=10,
mets_to_plot=["cpd03959_e0", "cpd00027_e0"],
title="Atrazine " + md)
# 2.1) add models, with random biomass
cons.add_model(mod_dir + "Arthrobacter_CORRECTED.json",
float(ar),
solver="glpk",
method="fba")
cons.add_model(mod_dir + "Halobacillus_sp_CORRECTED.json",
float(hb),
solver="glpk",
method="fba")
cons.add_model(mod_dir + "Halomonas_stevensii_CORRECTED.json",
float(hl),
solver="glpk",
method="fba")
#Se mete el ID de las bacterias. Si recorremos cons.models, ahí está el nombre de cada modelo, Arthrobacter empieza por k,
#por eso no nos interesa almacenarlo. Solo almacenamos los otros dos, que serán los que se añadan como perturbación.
st_ids = [id for id in cons.models if not id.startswith("k")]
print(f"Models were loaded on {md}.")
# 2.2) define initial medium => medium or exudate, already in JSON file
#root = deepcopy(gen_media[1])
del (gen_media[0]) # choose medium or exudate
del (gen_media[1])
del (gen_media[1])
# 2.3) add several random perturbations : strain | both | none
fix_biomass = 0.000034 # BGJ: fix biomass to add in the perturbation as probiotics
for pert in range(
3
): # BGJ: 2019.11.08 #SI QUEREMOS CAMBIAR EL NÚMERO DE PERTURBACIONES POR SIMULACIÓN, ES AQUÍ.
# perturbations always carry atrazine
gen_media.append({
"MEDIA": {
"cpd03959_e0": 0.15
},
"PERTURBATION": "ATRAZINE"
}) #Metemos una nueva entrada al diccionario
chosen = random.random()
if chosen > 0.75:
gen_media[-1]["MEDIA"][st_ids[
0]] = fix_biomass * 100 #Esto lo que hace es meter una nueva entrada en el diccionario dentro de "MEDIA"
#Con ese id y ese valor. Esta entrada se mete al últmo diccionario de gen_media que es lo último que appendeamos.
gen_media[-1]["PERTURBATION"] = st_ids[
0] #Nos referimos al último diccionario, al key "PERTURBATION" y lo cambiamos por ese valor.
elif chosen > 0.5: # 2019.11.12: BGJ: to increase biomass 2nd strain: st_ids[1]
gen_media[-1]["MEDIA"][st_ids[1]] = fix_biomass * 100
gen_media[-1]["PERTURBATION"] = st_ids[1]
elif chosen > 0.25:
gen_media[-1]["PERTURBATION"] = st_ids[0] + "_" + st_ids[1]
gen_media[-1]["MEDIA"][st_ids[0]] = fix_biomass * 100
gen_media[-1]["MEDIA"][st_ids[1]] = fix_biomass * 100
# elif chosen > 0.2:
# gen_media[-1]["MEDIA"] = root["MEDIA"]
# gen_media[-1]["PERTURBATION"] = "ROOT_EXUDATE"
else:
gen_media[-1]["MEDIA"]['cpd00009_e0'] = 1
gen_media[-1]["PERTURBATION"] = 'PHOSPHATE'
#else: none perturbation (only atrazine)
# 2.3) add 2nd perturbation (nothing)
# we need this to evaluate the last state of the consortium because the
# reward function in MDPbiome for this particular case evaluates degradation of
# atrazine 1 h after the simulation
gen_media.append({
"MEDIA": {
"cpd03959_e0": 0.15
},
"PERTURBATION": "ATRAZINE"
})
# 3) Write log
lock.acquire()
log(
cons, gen_media
) #Escribe en un archivo el objeto cons (nuestro experimento), junto con gen_media (están descritas las perturbaciones)
lock.release()
it = 1
t_pers = []
pers = []
for mper in gen_media:
if it == 1:
# 4) instantiate media
cons.media = cons.set_media(
mper["MEDIA"],
True) #El primer diccionario es el medio, lo instanciamos.
else:
for k in mper[
"MEDIA"]: # not needed at all... #Realizamos cambios en el medio.
if mper["MEDIA"][
k] == 0: #Si no hay de un metabolito, cambiamos su valor en el cons.media
cons.media[k] = 0
cons.add_mets(mper["MEDIA"],
True) #El resto de metabolitos los añadimos
pers.append(
mper["PERTURBATION"]
) #Metemos en esa lista el valor de la perturbación al final de cada vuelta de bucle.
# 5) run it
t_pers.append(cons.T[-1])
cons.run(verbose=False,
plot=False,
maxT=intervl + cons.T[-1],
integrator="FEA",
stepChoiceLevel=(0.00027, 0.5, 100000.),
outp=f'{md}/{md}_plot.png',
outf=f'{md}/plot.tsv')
it += 1
txpers = {
t_pers[i]: pers[i]
for i in range(len(t_pers))
} #creamos un diccionario een el que se guardan las perturbaciones en el orden que suceden (no termino de entenderlo)
# 6. Save simulation
with open(f'{md}/cons.p', 'wb') as f:
pickle.dump(cons, f) #El pickle es para guardar objetos en ficheros
with open(f'{md}/txpers.p', 'wb') as f:
pickle.dump(txpers, f)
tsv_filter(
f'{md}/plot.tsv', f'{md}/fluxes.tsv', txpers, inplace=False, v=cons.v
) #plot.tsv es el output de correr cons. fluxes.tsv se crea al instanciar Consortium.
if os.path.isfile(f'{md}/fluxes_filtered.tsv'):
os.unlink(f'{md}/fluxes.tsv')
mmodes.vis.plot_comm(cons)
lock.acquire()
print(f"\033[1;32;40mProcess with directory {md} out!\033[0m")
lock.release()
del (cons)
del (txpers)
del (gen_media)
return
def runn(md, all_random=False):
global lock
lock.acquire()
print("Starting simulation on", md)
lock.release()
subprocess.run(["mkdir", md])
subprocess.run(["cp", "-r", "fb1/ModelsInput", md])
path_curr = os.getcwd()
os.chdir(md)
mod_file = "ModelsInput"
intervl = 8
files = [
f for f in os.listdir(".")
if f.find("log_template.txt") != -1 or f[-3:] == "tsv"
]
subprocess.run(["rm"] + files)
with open(mod_file + "/media.json") as json_file:
gen_media = json.load(json_file)
inulin = {"PERTURBATION": "Inulin", "MEDIA": {"Inulin_29FruGlc": "0.0005"}}
fos = {"PERTURBATION": "FOS", "MEDIA": {"Kestose_C18H32O16": "0.005"}}
starch = {"PERTURBATION": "Starch", "MEDIA": {"Starch_11Glc": "0.00128"}}
faepraa2165 = {
"PERTURBATION": "Pro-Fprausx3.5",
"MEDIA": {
"FAEPRAA2165": "0.00035"
}
}
media = [gen_media[0]]
pers_perm = [inulin, fos, starch]
if all_random:
for i in range(0, 5):
if random.random() > 0.75:
pertadd = dcp(faepraa2165)
rand = [0.00001, 0.0001]
else:
pertadd = dcp(random.choice(pers_perm))
if pertadd == fos:
rand = [0.005, 0.015]
elif pertadd == inulin:
rand = [0.0002, 0.0006]
else:
rand = [0.0005, 0.0015]
perval = random.uniform(rand[0], rand[1])
pertadd["MEDIA"][list(pertadd["MEDIA"])[0]] = perval
pertadd["PERTURBATION"] += str(round(perval, 5))
media.append(pertadd)
else:
for i in range(0, 5):
if random.random() > 0.9:
media.append(faepraa2165)
else:
media.append(random.choice(pers_perm))
fp = random.uniform(0.0001, 0.00055)
ba = random.uniform(0.0001, 0.00055)
# 1) instantiate Consortium
volume_petri = 2 * 3.141593 * 45 * 45 * 15 * 1e-6 # 2pi*r²*h -> mm³ to L
cons = mmodes.Consortium(
stcut=-10,
mets_to_plot=["Kestose_C18H32O16", "Inulin_29FruGlc", "Starch_11Glc"],
v=volume_petri,
manifest="COMETS_manifest.txt",
work_based_on="name",
max_growth=10,
title=md)
# 2) add models
cons.add_model(mod_file + "/iFap484.V01.00.xml",
float(fp),
solver="glpk",
method="fba")
cons.add_model(mod_file + "/iBif452.V01.00.xml",
float(ba),
solver="glpk",
method="fba") # dMets = {glc.id: glc})
lock.acquire()
log(cons, media)
lock.release()
it = 1
t_pers = []
pers = []
for mper in media:
if it == 1:
# 4) instantiate media
cons.media = cons.set_media(mper["MEDIA"], True)
else:
# 4.2) or add perturbation
for k in mper["MEDIA"]:
if mper["MEDIA"][k] == 0:
cons.media[k] = 0
cons.add_mets(mper["MEDIA"], True)
pers.append(mper["PERTURBATION"])
# 5) run it
t_pers.append(cons.T[-1])
cons.run(verbose=False,
plot=False,
maxT=intervl + cons.T[-1],
integrator="vode",
stepChoiceLevel=(0., 0.5, 100000.),
outp=md + "_plot.png")
it += 1
txpers = {t_pers[i]: pers[i] for i in range(len(t_pers))}
# 6. Save relevant objects and generate output files
pickle.dump(cons, open("cons.p", "wb"))
pickle.dump(txpers, open("txpers.p", "wb"))
pickle.dump(media, open("media.p", "wb"))
#mmodes.vis.plot_comm(cons)
gen_biomass("plot.tsv", cons.v, cons.T[:-1], txpers)
gen_fluxes("flux_log_template.txt", cons.T, cons.manifest.models)
os.chdir(path_curr)
lock.acquire()
print(f"\033[1;32;40mProcess with directory {md} out!\033[0m")
del (cons)
del (txpers)
del (media)
lock.release()
return