dNdS Simulator uses R to simulate a stochastic branching process with a set of genotypes and phenotypes per cell. The main advantage of the model is that simulates immunoediting by introducing immunogenic and escape mutations as well as driver and passenger mutations.
We must first clone our repository, load libraries and functions for the simulation:
git clone https://github.com/luisgls/dNdSSimulator.git
library(ggpubr)
library(tidyverse)
library(crayon)
library(pio)
library(easypar)
library(parallel)
source('simulator_modelA.R')
source("functions_ZCsimulator.R")
source("simulator_plotting.R")
Note: Libraries pio and easypar can be obtained from https://github.com/caravagn/
Library ctree required for some function can be obtained using
install.packages("devtools")
devtools::install_github("caravagnalab/ctree")
Note: If you want to run the second model where the immune system attacks a population of immunogenic cells, you must source simulator_modelB.R instead of simulator_modelA.R.
As an example, we will run the simulator using the following paramters:
######### Number of simulations that we will perform #########
TASKS = 1
######## Define a Null set of parameters for our evolutionary model ###########
params = NULL
####Define initial population size (number of wild-type cells)
params['n0'] <- 5
######## Global parameter to stop simulation after N generations #########
params['GENERATIONS'] = 50
######## Global parameter to stop simulation after K number of alive cells #########
params['Kcapacity'] = 500
####### Set probability of cell survival #########
params['ps'] <- 0.5
params['pdiff'] <- (1 - params['ps'])
###### Set mutation rate parameter, the final mutation rate value used is equal to a combination of the following terms: u*C
params['u'] = 1 / (10 ^ 6) ### muts/ division / bp, mutation occurrence/ polymerase error
params['L'] = 50 * (10 ^ 6) ### Length of the genome/exome size of exome syn plus nonsyn mutations
params['C'] = 1 / (10 ^ 2) ### damage repair correction efficiency, it allows only 1 out of 100 number of errors to be preserved at each cell division
#######Set mutation probabilities
params['pnad'] <- 0.01 #Nonsyn - driver mutations (increase proliferation rate, oncogenes/ reduce correction efficiency, mut_modifiers, decrease death rate, etc)
params['pnai'] <- 0.10 #Nonsyn - deleterious mutations (immunogenic mutations)
params['pnak'] <- 0.00 #Nonsyn - deleterious mutations (death)
params['pnae'] <- 0.01 #Nonsyn - escape immune system mutations
params['pnsi'] <- params['pnai']/3 # Synonymous mutations in immunopeptidome
params['pnsd'] <- params['pnad']/3 # Synonymous mutations in driver
params['pnap'] <- 0.75 - ( params['pnad'] + params['pnai'] + params['pnae'] + params['pnak']) #Nonsyn - passenger mutations
params['pns'] <- 0.25 - ( params['pnsi'] + params['pnsd']) #Synonymous mutations
####Set value for the probability of the immune system to kill an immungenic population (PIs in the manuscript)
params['pattack'] = 0.5
Now that we have defined the parameters for the simulation, we will run one simulation using the command simulator:
M<-simulator(params)
The output contains the following information for each single cell:
id #cell id
pnad, pnap, pnai, pnae, pnak, pns, pnsi, pnsd #vector of probabilities to acquire a genotype
gnad, gnap, gnai, gnae, gnak, gns, gnsi, gnsd #vector of genotype
strategy # Death (never replicated), Survive (Replicate), immunogenic or escape cell
parent # parental cell ID
alive # status at the end of simulation
t_0 # current time
t_e # effective time
dnds # global dN/dS of the cell
dnds_i # immune dN/dS of the cell
dnds_d # driver dN/dS of the cel
psurv # Probability of effective cell division (delta)
pimmune # Probability of immune recognition
clone_id # Clone id assigned by driver status
clone_I_id # Clone id assigned by immunogenic status
parent_clone_id # Parental clone ID (driver status)
We have implemented the following code to run multiple simulations in parallel
######### Number of simulations that we will perform #########
TASKS = 1000
PARAMS_RUN1 = lapply(1:TASKS,function(w) list(params))
RUNS = run(
FUN = simulator,
PARAMS = PARAMS_RUN1,
silent = FALSE,
parallel = TRUE,
packages = c("ggpubr","tidyverse","ggmuller","crayon","pio")
)
This will create a list of objects, each object corresponds to one simulation.
We can also plot the results using the following function (after loading some libraries):
###Libraries for visualizing
library(reshape2)
library(tidygraph)
library(ggraph)
library(ggstatsplot)
library(ggmuller)
###Code to visualize results
# Visualization
if (count(M %>% dplyr::select(clone_id) %>% unique()) > 1) {
p1 <-
mullerplot(
M,
t_from = 1,
# t_to = 50,
legend = TRUE,
ps_0 = ps,
cutoff = .05,
add_clone_tree = FALSE
)
}
p1
Wich produces the following figure https://github.com/luisgls/dNdSSimulator/blob/master/dNdSTutorial_files/figure-html/visualize-1.png
The code shown above can also be accessed here:
If you want to use the simulator please cite the following article:
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent
