Fit clonal and subclonal components of the model to the residuals of the power-law model

Fit clonal and subclonal components of the model to the residuals of the power-law model

Usage

fit_subclones(
  object,
  N = 1:3,
  powerlaw_model_name = active_models(object),
  snvs_name = default_SNVs(object),
  cnvs_name = default_CNVs(object),
  method = "BMix",
  upper_f_limit = 0.75,
  clip_sif = NULL,
  clip_input = file.path(tempdir(), "clip_input"),
  clip_output = file.path(tempdir(), "clip_output"),
  verbose = get_cevomod_verbosity()
)

fit_subclones_bmix(
  object,
  N = 1:3,
  powerlaw_model_name = active_models(object),
  snvs_name = default_SNVs(object),
  upper_f_limit = 0.75,
  verbose = get_cevomod_verbosity()
)

fit_subclones_clip(
  object,
  powerlaw_model_name = active_models(object),
  snvs_name = default_SNVs(object),
  cnvs_name = default_CNVs(object),
  upper_f_limit = 0.75,
  clip_sif = NULL,
  clip_input = file.path(tempdir(), "clip_input"),
  clip_output = file.path(tempdir(), "clip_output"),
  verbose = get_cevomod_verbosity()
)

fit_subclones_mclust(
  object,
  N = 1:3,
  powerlaw_model_name = active_models(object),
  snvs_name = default_SNVs(object),
  upper_f_limit = 0.75,
  verbose = get_cevomod_verbosity()
)

Arguments

object

cevodata object

N

Vector of numbers of clones to for models

powerlaw_model_name

Residual of which powerlaw model to use? powerlaw_fixed/powerlaw_optim

snvs_name

Which snvs to to use?

cnvs_name

Which cnvs to to use?

method

Clustering method to use. Currently supported methods:

• mclust - the fastest method, approximately 3-4 times faster than BMix, but uses a gaussian mixture modelling

• BMix - is more accurate, considers subclones as binomial clusters, slightly slower

• CliP - Clonal structure identification through penalizing pairwise differences

upper_f_limit

ignore variants with f higher than

clip_sif

Apptainer image file with CliP. If NULL, cevomod seeks for the CliP.sif file in the containers_dir (if set before) and in the current working directory. The container should be build with build_clip_container().

clip_input

Path to store the CliP input files

clip_output

Path to store the CliP output files

verbose

Verbose?

Functions

• fit_subclones(): Provides a common interface for all other methods, runs the selected method and passes all the required arguments down.

• fit_subclones_bmix(): Fit subclonal distributions to neutral model residuals using BMix

• fit_subclones_clip(): Fit subclonal distributions to neutral model residuals using CliP - Clonal structure identification through penalizing pairwise differences (github). Requires that the Apptainer installed and

  • the path to the CliP.sif container image is provided,

  • or CliP.sif exists in the containers_dir (set_containers_dir())

  • or CliP.sif exists in the current working directory. The container should be build with build_clip_container().

• fit_subclones_mclust(): Fit subclonal distributions to neutral model residuals using mclust

Examples

if (FALSE) {
# Using BMix
fit_subclones(test_data_fitted)
# or
fit_subclones_bmix(test_data_fitted)

# Using mclust
fit_subclones(test_data_fitted, method = "mclust")
# or
fit_subclones_mclust(test_data_fitted)

# Using CliP
set_containers_dir(selected_dir)
build_clip_container()
fit_subclones(test_data_fitted, method = "CliP")
# or
fit_subclones_clip(test_data_fitted)
}