ncclust {GeneNT} | R Documentation |
This function does network constrained clustering based on Floyd-Warshall algorithm, R function allshortestpaths()in R packpage e1071.
ncclust(p, pG2, kG2)
p |
p is the exponential tuning factor with default value 1. it can also be set to other intergers whenever necessary. |
pG2 |
pG2 is the gene pairs that are screened using the two-stage algorithm based on Pearson correlation statistic. |
kG2 |
kG2 is the gene pairs that are screened using the two-stage algorithm based on Kendall correlation statistic. |
This function is written in comparison to the traditional clustering implemented as tdclust().
This function returns a network constrained distance matrix that can be used by any distance based clustering software.
Dongxiao Zhu (http://dongxiaozhu.googlepages.com/home)
Zhu, D., Hero, A.O., Qin, Z.S. and Swaroop, A. High throughput screening of co-expressed gene pairs with controlled False Discovery Rate (FDR) and Minimum Acceptable Strength (MAS). J Comput Biol, in press. Zhu, D., Hero, A.O., Hong, C., Khanna, R., and Swaroop A. Network constrained clustering for gene microarray data. {it Submitted}
## load GeneNT library library(RBGL) library(e1071) library(GeneNT) ## load example dataset data(dat) ## Note, data matrix name has to be "dat" ## use (FDR, MAS) criteria (0.2, 0.5) as example to screen gene pairs ## use "BY" procedure to do multiple comparison adjustment g1 <- corfdrci(0.2, 0.5, "BY") pG1 <- g1$pG1 ## pG2 contains gene pairs that passed two-stage screening pG2 <- g1$pG2 ## use (FDR, MAS) criteria (0.2, 0.5) as example to screen gene pairs ## use "BY" procedure to do multiple comparison adjustment g2 <- kendallfdrci(0.2, 0.5, "BY") kG1 <- g2$kG1 ## kG2 contains gene pairs that passed two-stage screening kG2 <- g2$kG2 ## generate Pajek compatible matrix to visualize network getBM(pG2, kG2) ## clustering from network using network constraint clustering, for example, p = 6. ncclust(6, pG2, kG2)