fit_mt_models

Fits models for analyzing multi-timepoint perturbation data using a cell dataset (cds).
Allows specification of parameters for model formulae, perturbation settings, batch correction, and time interval handling.

fit_mt_models(
  cds,
  sample_group,
  cell_group,
  main_model_formula_str = NULL,
  num_time_breaks = 3,
  nuisance_model_formula_str = "~1",
  ctrl_ids = NULL,
  mt_ids = NULL,
  sparsity_factor = 1,
  vhat_method = "bootstrap",
  interval_col = "timepoint",
  perturbation_col = "knockout",
  batch_col = "expt",
  newdata = tibble::tibble(),
  start_time = NULL,
  stop_time = NULL,
  interval_step = 2,
  log_abund_detection_thresh = -5,
  q_val = 0.1,
  edge_allowlist = NULL,
  edge_denylist = NULL,
  keep_cds = TRUE,
  keep_ccs = TRUE,
  verbose = FALSE,
  num_threads = 1,
  backend = "nlopt",
  penalize_by_distance = TRUE,
  independent_spline_for_ko = TRUE,
  num_bootstraps = 10,
  embryo_size_factors = NULL,
  batches_excluded_from_assembly = c()
)

Arguments

• cds
  cell dataset
  The input dataset containing cell data.

• sample_group
  character
  Sample grouping variable.

• cell_group
  character
  Cell grouping variable.

• main_model_formula_str
  character
  Main model formula. Default is NULL.

• num_time_breaks
  integer
  Number of time breaks. Default is 3.

• nuisance_model_formula_str
  character
  Nuisance model formula. Default is "~1".

• ctrl_ids
  vector
  Control identifiers. Default is NULL.

• mt_ids
  vector
  Multi-timepoint identifiers. Default is NULL.

• sparsity_factor
  numeric
  Sparsity adjustment factor. Default is 1.

• vhat_method
  character
  Method for variance estimation. Default is "bootstrap".

• interval_col
  character
  Column name for time intervals. Default is "timepoint".

• perturbation_col
  character
  Column name for perturbation IDs. Default is "knockout".

• batch_col
  character
  Column name for batch identifiers. Default is "expt".

• newdata
  tibble
  New data for prediction. Default is an empty tibble.

• start_time
  numeric
  Start time. Default is NULL.

• stop_time
  numeric
  Stop time. Default is NULL.

• interval_step
  integer
  Step size for time intervals. Default is 2.

• log_abund_detection_thresh
  numeric
  Log abundance detection threshold. Default is -5.

• q_val
  numeric
  Q-value threshold for significance. Default is 0.1.

• edge_allowlist
  vector
  Allowed edges in the analysis. Default is NULL.

• edge_denylist
  vector
  Denied edges in the analysis. Default is NULL.

• keep_cds
  logical
  Whether to retain the CDS in output. Default is TRUE.

• keep_ccs
  logical
  Whether to retain the CCS in output. Default is TRUE.

• verbose
  logical
  Whether to print verbose output. Default is FALSE.

• num_threads
  integer
  Number of computation threads. Default is 1.

• backend
  character
  Optimization backend. Default is "nlopt".

• penalize_by_distance
  logical
  Whether to penalize by distance. Default is TRUE.

• independent_spline_for_ko
  logical
  Use independent splines for knockouts. Default is TRUE.

• num_bootstraps
  integer
  Number of bootstraps for variance estimation. Default is 10.

• embryo_size_factors
  named vector
  Embryo size factors. Default is NULL.

• batches_excluded_from_assembly
  vector
  Batch IDs to exclude. Default is an empty vector.

Value

A tibble containing fitted perturbation models and associated metadata.

Details

This function performs multi-timepoint perturbation analysis on cell count data.
It supports handling of batch effects, customizable formulas, and exclusion of specific data components.
User-supplied embryo size factors and batch exclusions are supported for fine-tuned control.

Examples

# Example usage:
fit_mt_models(
  cds,
  sample_group = "sample",
  cell_group = "cell_type"
)