Skip to contents

Description: OTTER infers gene regulatory networks using TF DNA binding motif (W), TF PPI (P), and gene coexpression (C) through minimzing the following objective: min f(W) with f(W) = (1-lambda)*||WW' - P||^2 + lambda*||W'W - C||^2 + (gamma/2)*||W||^2

Usage

otter(W, P, C, lambda = 0.035, gamma = 0.335, Iter = 60, eta = 1e-05, bexp = 1)

Arguments

W

: TF-gene regulatory network based on TF motifs as a matrix of size (t,g), g=number of genes, t=number of TFs

P

: TF-TF protein interaction network as a matrix of size (t,t)

C

: gene coexpression as a matrix of size (g,g)

lambda

: tuning parameter in [0,1] (higher gives more weight to C)

gamma

: regularization parameter

Iter

: number of iterations of the algorithm

eta

: learning rate

bexp

: exponent influencing learning rate (higher means smaller)

Outputs:

Value

W : Predicted TF-gene complete regulatory network as an adjacency matrix of size (t,g).

Details

Inputs:

Examples


W=matrix(rexp(100, rate=.1), ncol=10)
C=matrix(rexp(100, rate=.1), ncol=10)
P=matrix(rexp(100, rate=.1), ncol=10)

# Run OTTER algorithm
W <- otter(W, P, C)