Ivis4ml / FSSEM

Inference of Differential Gene Regulatory Networks Based on Gene Expression and Genetic Perturbation data.

FSSEM

FSSEM(Fused Sparse Structural Equational Model) is a package for Inference of Differential Gene Regulatory Networks Based on Gene Expression and Genetic Perturbation data.

Installation

Install R

source("./src/solver.R")

or

source("./src/solver.min.R")

Input format

For now, the fssem function multiSML_iPALM or genSML_iPALM requires data of gene expression and genetic perturbation data(eQTL data). multiSML_iPALM: run FSSEM if X1=X2; genSML_iPALM: run FSSEM if X1 != X2.

I use the list data for representing the info of all data. One data list has two elements:

• obs: Observation data for genes and eQTLs, it has:

  • Y1, Y2: a gene expression matrix under 2 conditions.
  • X or X1,X2: eQTL data for each gene under 2 conditions.
  • sk: nonzero eQTL index for each gene.

• var: Inherent variable of FSSEM, it has:

  • N: sample size per each condition.
  • Ng: total gene number.
  • Ne: total eQTL number.

Usage

source("./src/demo.R")

Run demo.R code in src folder. And in this demo, you can simulate gene expression and genetic perturbation data with random generated network structures. You can randomly generate networks with parameters:

• N: sample size per each condition;
• Ng: gene number;
• Ne: eQTL number;
• Ns: sparse ratio of gene network(number of nonzero entries per gene).
• sigma: noise standard deviation.