gcaer

Branch
`master`
`develop`

Work with GCAE from R.

GCAE GitHub repository

Installation

gcaer is not on CRAN yet. To install gcaer:

library(remotes)
install_github("richelbilderbeek/gcaer")

This assumes you have the remotes package installed.

Install GCAE versions

To install GCAE:

library(gcaer)
install_gcae()

Examples

Get the GCAE help text:

library(gcaer)
get_gcae_help_text()

Running GCAE

Run GCAE:

library(gcaer)
run_gcae("--help")

Full experiment

Instead of using the multiple steps by GenoCAE, do_gcae_experiment does all of these for you.

Here is an example of a full experiment:

# Create the parameters for the experiment
gcae_experiment_params <- create_gcae_experiment_params(
  gcae_options = create_gcae_options(),
  gcae_setup = create_test_gcae_setup(
    model_id = "M0",
    superpops = get_gcaer_filename("gcae_input_files_1_labels.csv"),
    pheno_model_id = "p0"
  ),
  analyse_epochs = c(1, 2),
  metrics = "f1_score_3,f1_score_5"
)

# Do the experiment
gcae_experiment_results <- do_gcae_experiment(
  gcae_experiment_params = gcae_experiment_params
)

# Save the experiment's results
save_gcae_experiment_results(
  gcae_experiment_results = gcae_experiment_results,
  folder_name = gcae_experiment_params$gcae_setup$trainedmodeldir
)

# Create the plots for the experiment's results
create_plots_from_gcae_experiment_results(
  folder_name = gcae_experiment_params$gcae_setup$trainedmodeldir
)

Workflow

To do the full GCAE workflow, a gcae_setup is needed, from which the respective gcae_[x] functions are called, where [x] matches the first GCAE CLI argument (for example, use gcaer's gcae_train to do the same as run_gcae.py train)

gcae_setup <- create_gcae_setup(
  datadir = file.path(get_gcae_folder(), "example_tiny/"),
  data = "issue_6_bin",
  model_id = "M1",
  pheno_model_id = "p2",
  superpops = file.path(datadir, "HO_superpopulations")
)

# 2. Train, approx 3 mins
train_filenames <- gcae_train(
  gcae_setup = gcae_setup,
  epochs = 3,
  save_interval = 1
)

# 3. Project
project_filenames <- gcae_project(
  gcae_setup = gcae_setup
)
project_results <- parse_project_files(project_filenames)

# 4. Evaluate
evaluate_filenames <- gcae_evaluate(
  gcae_setup,
  metrics = "f1_score_3,f1_score_5",
  epoch = 3
)

evaluate_results <- parse_evaluate_filenames(
  evaluate_filenames, 
  epoch = 3
)

bschilder / gcaer