compute.PFN.par: Parallelized PFN computation
In MEGENA: Multiscale Clustering of Geometrical Network

Description Usage Arguments Details Value Author(s)

PFN construction by parallelized edge screening.

1 2	compute.PFN.par(sortedEdge,Ng,maxENum,Njob,Ncore,max.skipEdges = NULL, keep.track = TRUE,initial.links = NULL)

`sortedEdge`	3-column matrix for the input edgelist (e.g. - correlation pair list). Must be sorted by third column, which is usually weight vector.
`Ng`	integer. number of genes included in sortedEdge.
`maxENum`	Maximum number of edges to include in final PFN. The theoretical maximal number enforced by Euler's formula is 3(Ng-2).
`max.skipEdges`	Maximum number of edges to be counted before any valid edge to be included in PFN. This works as a termination condition to avoid exhaustive planarity testing over all edges provided in sortedEdge.
`Njob`	Number of edges to be passed to each core for parallelized edge screening.
`Ncore`	Number of cores to utilize.
`keep.track`	TRUE/FALSE logical. Indicate if the record of PFN construction is saved in temporary file "pfg_el.RData". Default is TRUE.
`initial.links`	If provided, PFN construction will restart by regarding these initial.links as already-built PFN.

This is parallelized implementation of PFN construction, where it is possible to re-capture PFN construction by providing already computed edgelist into initial.links. Although provivded, this function itself may require careful caution and users are encouraged to use more user-friendly "calculate.PFN()" instead.

A 3-column matrices, where first two columns are integer indices for vertices, and third is the weight vector.

Won-Min Song

MEGENA documentation built on May 1, 2019, 8:07 p.m.