tigress: Trustful Inference of Gene REgulation using Stability...
In jpvert/tigress: Trustful Inference of Gene REgulation using Stability Selection

This function implements the TIGRESS model of Haury et al. for gene regulatory network (GRN) inference from gene expression data. Given a matrix of expression data, TIGRESS infers regulation from transcription factor (TFs) to any target gene by performing a regression from the TF expression data to the target gene expression, and scoring the candidate TFs by a stability selection (SS) score.

1
2
3

tigress(expdata, tflist = colnames(expdata), targetlist = colnames(expdata),
  alpha = 0.2, nstepsLARS = 5, nsplit = 100, normalizeexp = TRUE,
  scoring = "area", allsteps = TRUE, verb = FALSE, usemulticore = FALSE)

`expdata`	A matrix of expression. Each row is an experiment, each column a gene. The gene names are the column names.
`tflist`	The list of TFs names. The TF names should match the names in the expression matrix (default: all genes are considered TFs).
`targetlist`	The list of targets' names. The targets' names should match the names in the expression matrix (default: all genes are considered targets).
`alpha`	The `alpha` parameter for randomization in stability selection. Should be between 0 and 1 (default `0.2`).
`nstepsLARS`	Number of LARS steps to perform in stability selection (default `5`).
`nsplit`	Number of sample splits to perform in stability selection (default `100`)
`normalizeexp`	A boolean indicating whether we should mean center and scale to unit variance the expression data for each gene (default `TRUE`).
`scoring`	The method for scoring a feature in stability selection. If `"area"`, the score is the area under the stability curve up to nstepsLARS steps, as proposed by Haury et al. If `"max"`, the score is the maximum value of the stability curve, as proposed by Meinshausen and Bühlmann in the original paper (default `"area"`).
`allsteps`	A boolean indicating whether we should output the solutions for all values of LARS steps up to nstepsLARS, or only for nstepsLARS. It does not cost more computation to compute all solutions (default `TRUE`).
`verb`	A boolean indicating verbose mode. If `TRUE`, print messages about what we are doing, otherwise remain silent (default `FALSE`).
`usemulticore`	A boolean indicating whether multicore parallel computing is used. This requires the package `parallel` (default `FALSE`).

A list of matrices (or a single matrix if allsteps=FALSE) with the scores of each TF x target candidate interaction. Each row corresponds to a TF, each column to a target.

jpvert/tigress documentation built on May 3, 2019, 4:04 p.m.