Fit sub-MRF ensemble for a multi-omics connection list

Higher-level wrapper that matches the interface of fit_multi_forest(). For each directed connection (response_block -> predictor_block), features are sub-sampled from both blocks independently, and the resulting n x n matrices are averaged across replicates.

Usage

fit_sub_multi_rfsrc(
  dat.list,
  connect_list,
  n_sub = 15L,
  frac_response = 0.2,
  frac_predictor = 0.2,
  ntree_per_sub = 25L,
  mtry = NULL,
  ytry = NULL,
  min_response = 20L,
  min_predictor = 30L,
  min_response_for_sub = 500L,
  min_predictor_for_sub = 500L,
  ntree_full = 300L,
  enhanced = FALSE,
  compute_imd = FALSE,
  imd_args = list(),
  seed = 529L,
  parallel = FALSE,
  cores = 2L,
  parallel_connections = FALSE,
  cores_connections = NULL,
  verbose = TRUE,
  ...
)

Arguments

dat.list: Named list of data frames (samples x features).
connect_list: A list of connections, each a character vector of length 2: c(response_name, predictor_name). Same format as used by fit_multi_forest().
n_sub: Integer; number of sub-MRF replicates to fit.
frac_response: Fraction of response columns to sample per replicate (default 0.2). Convergence analysis shows symmetric low fractions (0.1–0.3 for both response and predictor) yield the best approximation of the full forest-weight matrix.
frac_predictor: Fraction of predictor columns to sample per replicate (default 0.2). Avoid setting this to 1.0 when frac_response is small, as the asymmetric case leads to poor convergence (each sub-model overfits to its response subset).
ntree_per_sub: Number of trees per sub-MRF (default 25 for direct calls; the workflow inherits the main tree budget by using ceiling(ntree / n_sub) unless explicitly overridden).
mtry: Number of candidate X variables per split. Passed through to fit_forest().
ytry: Number of candidate Y variables per split (0 = min(qy, ceiling(p/3)) default).
min_response: Minimum number of response columns per sub-MRF.
min_predictor: Minimum number of predictor columns per sub-MRF.
min_response_for_sub: Minimum response-block size below which all response features are used instead of sub-sampling.
min_predictor_for_sub: Minimum predictor-block size below which all predictor features are used instead of sub-sampling.
ntree_full: Number of trees used when both blocks are below their sub-sampling thresholds and a full forest is fitted instead.
enhanced: Logical; if TRUE, compute soft enhanced proximity using full-data sample embeddings within each sub-model. The result is stored as $enhanced_prox in the output so that mrf3_cl_prox() can use it directly without re-traversing trees. Default FALSE because the extra computation is non-trivial.
compute_imd: Logical; whether to pre-compute IMD weights across sub-models.
imd_args: Named list of arguments passed to get_imp_forest() when compute_imd = TRUE.
seed: Base random seed.
parallel: Logical; if TRUE, fit replicates in fresh PSOCK worker processes.
cores: Number of cores used for within-connection parallelism. The default for this wrapper is 2L.
parallel_connections: Logical; whether distinct directed connections may be fitted in fresh PSOCK worker processes.
cores_connections: Optional total core budget for connection-level parallelism.
verbose: Logical; print progress messages.
...: Passed to fit_sub_mrf() and then fit_forest().

Value

A named list of rfsrc-compatible objects, one per connection. Names follow the "response_predictor" convention used by fit_multi_forest(). Each object contains forest.wt, forest.wt.oob, proximity, xvar, yvar.

Details

Sub-MRF uses multiRF for its internal fits. OOB forest-weight reconstruction uses the returned inbag matrix.