dyngen: A multi-modal simulator for spearheading single-cell omics analyses

A toolkit for generating synthetic single cell data.

Step 1, initialise dyngen model

initialise_model(): Define and store settings for all following steps. See each of the sections below for more information.
Use a predefined backbone:
Create a custom backbone:
- backbone()
- bblego()
- bblego_linear()
- bblego_branching()
- bblego_start()
- bblego_end()
Visualise the backbone:
- plot_backbone_modulenet()
- plot_backbone_statenet()

Step 2, generate TF network

generate_tf_network(): Generate a transcription factor network from the backbone
tf_network_default(): Parameters for configuring this step

Step 3, add more genes to the gene network

generate_feature_network(): Generate a target network
feature_network_default(): Parameters for configuring this step
plot_feature_network(): Visualise the gene network

Step 4, generate gene kinetics

generate_kinetics(): Generate the gene kinetics
kinetics_default(), kinetics_random_distributions(): Parameters for configuring this step

Step 5, simulate the gold standard

generate_gold_standard(): Simulate the gold standard backbone, used for mapping to cell states afterwards
gold_standard_default(): Parameters for configuring this step
plot_gold_mappings(): Visualise the mapping of the simulations to the gold standard
plot_gold_simulations(): Visualise the gold standard simulations using the dimred
plot_gold_expression(): Visualise the expression of the gold standard over simulation time

Step 6, simulate the cells

generate_cells(): Simulate the cells based on its GRN
simulation_default(): Parameters for configuring this step
simulation_type_wild_type(), simulation_type_knockdown(): Used for configuring the type of simulation
kinetics_noise_none(), kinetics_noise_simple(): Different kinetics randomisers to apply to each simulation
plot_simulations(): Visualise the simulations using the dimred
plot_simulation_expression(): Visualise the expression of the simulations over simulation time

Step 7, simulate cell and transcripting sampling

generate_experiment(): Sample cells and transcripts from experiment
list_experiment_samplers(), experiment_snapshot(), experiment_synchronised(): Parameters for configuring this step
simtime_from_backbone(): Determine the simulation time from the backbone
plot_experiment_dimred(): Plot a dimensionality reduction of the final dataset

Step 8, convert to dataset

as_dyno(), wrap_dataset(): Convert a dyngen model to a dyno dataset
as_anndata(): Convert a dyngen model to an anndata dataset
as_sce(): Convert a dyngen model to a SingleCellExperiment dataset
as_seurat(): Convert a dyngen model to a Seurat dataset

One-shot function

generate_dataset(): Run through steps 2 to 8 with a single function
plot_summary(): Plot a summary of all dyngen simulation steps

Data objects

example_model: A (very) small toy dyngen model, used for documentation and testing purposes
realcounts: A set of real single-cell expression datasets, to be used as reference datasets
realnets: A set of real gene regulatory networks, to be sampled in step 3

Varia functions

dyngen: This help page
get_timings(): Extract execution timings for each of the dyngen steps
combine_models(): Combine multiple dyngen models
rnorm_bounded(): A bounded version of rnorm()
runif_subrange(): A subrange version of runif()

Examples

model <- initialise_model(
  backbone = backbone_bifurcating()
)
# \dontshow{
# actually use a smaller example 
# to reduce execution time during
# testing of the examples
model <- initialise_model(
  backbone = model$backbone,
  num_cells = 5,
  num_targets = 0,
  num_hks = 0,
  gold_standard_params = gold_standard_default(census_interval = 1, tau = 0.1),
  simulation_params = simulation_default(
    burn_time = 10,
    total_time = 10,
    census_interval = 1,
    ssa_algorithm = ssa_etl(tau = 0.1),
    experiment_params = simulation_type_wild_type(num_simulations = 1)
  )
)
# }
# \donttest{
model <- model %>%
  generate_tf_network() %>%
  generate_feature_network() %>%
  generate_kinetics() %>%
  generate_gold_standard() %>%
  generate_cells() %>%
  generate_experiment()
#> Generating TF network
#> Sampling feature network from real network
#> Generating kinetics for 33 features
#> Generating formulae
#> Generating gold standard mod changes
#> Precompiling reactions for gold standard
#> Running gold simulations
#> 
  |                                                  | 0 % elapsed=00s   
  |========                                          | 14% elapsed=00s, remaining~00s
  |===============                                   | 29% elapsed=00s, remaining~00s
  |======================                            | 43% elapsed=00s, remaining~00s
  |=============================                     | 57% elapsed=00s, remaining~00s
  |====================================              | 71% elapsed=00s, remaining~00s
  |===========================================       | 86% elapsed=00s, remaining~00s
  |==================================================| 100% elapsed=01s, remaining~00s
#> Precompiling reactions for simulations
#> Running 1 simulations
#> Mapping simulations to gold standard
#> Warning: Simulation does not contain all gold standard edges. This simulation likely suffers from bad kinetics; choose a different seed and rerun.
#> Performing dimred
#> Simulating experiment
#> Warning: Certain backbone segments are not covered by any of the simulations. If this is intentional, please ignore this warning.
#>   Otherwise, increase the number of simulations (see `?simulation_default`) or decreasing the census interval (see `?simulation_default`).
  
dataset <- wrap_dataset(model, format = "dyno")
# format can also be set to "sce", "seurat", "anndata" or "list"

# library(dynplot)
# plot_dimred(dataset)
# }