findBF: RFRE and Boolean function reconstruction

In ningshi/ATEN: ATEN

Description

A recursive feature reconstruction and elimination (RFRE) procedure for selecting a set of important prime implicants as features. Then final solution (i.e. Boolean function) is generated.

Usage

findBF(B, PIs, target, parameters, datalist, datasamples, seed)

Arguments

B	The number of And/Or trees that used for generating a set of prime
PIs	A list of prime implicants
target	The integer indicating which node is the target node
parameters	A vector includes 6 parameters; startT, endT and maxIter refer to the upper, lower temperature (on a log10 scale) and the maximum number of iterations used in simulated annealing algorithm maxK represents the maximum number of input nodes of the target node (the size of the tree), it is required when prior knowledge, e.g. the in-degree is 8, is available. If such information is not known, then it can be set as a very large value, e.g. '.Machine$integer.max' rate represents how many non-important PIs are removed in each recursion. nodes represents the number of node in the Boolean network
datalist	A data list generated by buildTimeSeries or bootstrap
datasamples	The number of nodes in the Boolean network
seed	This seed is made for parLapply to reproduce the results

Value

A final And/Or tree (i.e., a Boolean function)

See Also

findPIs

Examples

## After we get a set of prime implicants, we use this function to generate the final Boolean function
## choose the same target gene
target<-3
datalist[[2]]<-generateData(PIs,datalist)
datalist[[3]]<-matrix(datalist[[3]][,target])
datasamples<-bootstrap(datalist)
datasamples$respinbag<-matrix(datasamples$respinbag)
datasamples$respoutbag<-matrix(datasamples$respoutbag)

parameters<-c(startT=2,endT=-1,maxIter=2000,maxK=8,rate=0.2,nodes=10)

## Identify the final Boolean function with 5 trees in the forest
findBF(5,PIs,target,parameters,datalist,datasamples,123)

ningshi/ATEN documentation built on April 27, 2021, 7:40 a.m.