GEARS

require(GCModeller);

#' GEARS: the graph neural network based in silico perturbation prediction toolkit
imports "GEARS" from "biosystem";

GEARS: the graph neural network based in silico perturbation prediction toolkit

This R# package module provides the toolkit for train a GEARS model(Gene Expression Additive Response Simulator) from the Perturb-seq experiment data, the trained model can be used for predict the gene expression response of the in silico gene perturbation(the knockout/overexpression/knockdown experiment):

• new: create a new GEARS model from the gene expression matrix, the prior regulatory network and the model configuration;
• training_set: set the Perturb-seq training sample set of the GEARS model;
• train: train the GEARS model with the given training sample set.

the trained GEARS model object implements the InsilicoPerturbationExperiment interface, so that it can be used by the knockouts, overexpress and knockdown api of the bnlearn package module, and the perturbation result can be exported via the make_exports api.