hitting.time: —————— Alternative approach to prioritize genes

In matthiasheinig/QTLnetwork: Basic package to apply random walks for QTL candidate gene priorization

Description

We can compute expected hitting times for the random walks instead of using the propagation approach.

Usage

hitting.time(Asparse, n.eigs = 20, from = NULL, to = NULL,
  sum = "none")

Arguments

n.eigs	The number of eigenvectors used for the approximation. If this is smaller than 2, the pseudo inverse will be computed
from	Source nodes for which we need the transition probabilties.
to	Target nodes for which we need the transition probabilties.#'
sum	Can be "none", "from", "to" or "both" and will sum up the propagation matrix over one or the other or both lists.

Details

Using from and to nodes avoids to compute the full propagation matrix which is too large to fit in memory. When from and to are NULL the full matrix will be computed. For sum, "none" returns a "from" by "to" matrix; "from" returns a vector of length nrow(Asparse); "to" returns a vector of length nrow(Asparse) and "both" returns a nrow(Asparse) by 2 matrix with the from and to vectors. The i-th component of the "from sum" represents the average hitting time going from node i to any of the "to" nodes. The i-th component of the "to sum" represents the average hitting time going from any "from" node to node i

Value

Table of hitting times

Author(s)

Matthias Heinig <[e-mail]>

—————————————————————————–

matthiasheinig/QTLnetwork documentation built on June 29, 2021, 10:11 p.m.