| ... | ... |
@@ -22,7 +22,7 @@ Depends: R (>= 2.10), methods, BiocGenerics (>= 0.13.8), Biobase, plyr, |
| 22 | 22 |
Imports: AnnotationDbi, lumi, fpc, lumiHumanAll.db |
| 23 | 23 |
Suggests: biomaRt, GOstats, cluster, hgu133a.db, mclust, igraph, |
| 24 | 24 |
breastCancerMAINZ, breastCancerTRANSBIG, breastCancerUPP, |
| 25 |
- breastCancerVDX |
|
| 25 |
+ breastCancerVDX, future, future.apply |
|
| 26 | 26 |
Enhances: doMC |
| 27 | 27 |
Collate: 'AllClasses.R' 'AllGeneric.R' 'methods-IcaSet.R' |
| 28 | 28 |
'methods-MineICAParams.R' 'compareAnalysis.R' |
| ... | ... |
@@ -105,27 +105,60 @@ clusterFastICARuns <- function(X, nbComp, nbIt=100, alg.type = c("deflation", "p
|
| 105 | 105 |
## compute Iq indices and extract centrotypes |
| 106 | 106 |
# Iq=avg(intra-cluster similarity) - avg(extra-cluster similarity) |
| 107 | 107 |
|
| 108 |
- Iq <- |
|
| 109 |
- foreach(clus=unique(partition), .combine=c) %dopar% {
|
|
| 110 |
- indC <- which(partition==clus) |
|
| 111 |
- if (length(indC)>1) {
|
|
| 112 |
- if (funClus != "pam") |
|
| 113 |
- centrotypes <- c(centrotypes,indC[which.max(apply(sim[indC,indC],1,sum))]) |
|
| 114 |
- internalSim <- mean(sim[indC,indC]) |
|
| 115 |
- } |
|
| 116 |
- else {
|
|
| 117 |
- if (funClus != "pam") |
|
| 118 |
- centrotypes <- c(centrotypes,indC) |
|
| 119 |
- internalSim <- sim[indC,indC] |
|
| 120 |
- } |
|
| 121 |
- externalSim <- mean(sim[indC,setdiff(1:ncol(sim),indC)]) |
|
| 122 |
- iq <- internalSim-externalSim |
|
| 123 |
- |
|
| 124 |
- return(iq) |
|
| 125 |
- } |
|
| 108 |
+ getIqCentr <- function(clus, partition, sim, funClus) {
|
|
| 109 |
+ indC <- which(partition==clus) |
|
| 110 |
+ if (length(indC)>1) {
|
|
| 111 |
+ if (funClus != "pam") |
|
| 112 |
+ centrotypes <- indC[which.max(apply(sim[indC,indC],1,sum))] |
|
| 113 |
+ internalSim <- mean(sim[indC,indC]) |
|
| 114 |
+ } else {
|
|
| 115 |
+ if (funClus != "pam") |
|
| 116 |
+ centrotypes <- indC |
|
| 117 |
+ internalSim <- sim[indC,indC] |
|
| 118 |
+ } |
|
| 119 |
+ externalSim <- mean(sim[indC,setdiff(1:ncol(sim),indC)]) |
|
| 120 |
+ iq <- internalSim-externalSim |
|
| 121 |
+ |
|
| 122 |
+ return(c(centrotype=centrotypes,iq=iq)) |
|
| 123 |
+ } |
|
| 124 |
+ |
|
| 125 |
+ |
|
| 126 |
+ if (requireNamespace("future", quietly = TRUE) & requireNamespace("future.apply", quietly = TRUE)) {
|
|
| 127 |
+ future::plan(future::multiprocess) ## => parallelize on your local computer |
|
| 128 |
+ iqcentr <- |
|
| 129 |
+ future.apply::future_lapply(unique(partition), getIqCentr, partition=partition, sim=sim, funClus=funClus) |
|
| 130 |
+ } else {
|
|
| 131 |
+ iqcentr <- |
|
| 132 |
+ lapply(unique(partition), getIqCentr, partition=partition, sim=sim, funClus=funClus) |
|
| 133 |
+ } |
|
| 134 |
+ |
|
| 135 |
+ iqcentr <- do.call(rbind, iqcentr) |
|
| 136 |
+ Iq <- iqcentr[,"iq"] |
|
| 137 |
+ if (funClus != "pam") centrotypes <- iqcentr[,"centrotype"] |
|
| 138 |
+ |
|
| 139 |
+ |
|
| 140 |
+ |
|
| 141 |
+# Iq <- |
|
| 142 |
+# foreach(clus=unique(partition), .combine=c) %dopar% {
|
|
| 143 |
+# indC <- which(partition==clus) |
|
| 144 |
+# if (length(indC)>1) {
|
|
| 145 |
+# if (funClus != "pam") |
|
| 146 |
+# centrotypes <- c(centrotypes,indC[which.max(apply(sim[indC,indC],1,sum))]) |
|
| 147 |
+# internalSim <- mean(sim[indC,indC]) |
|
| 148 |
+# } |
|
| 149 |
+# else {
|
|
| 150 |
+# if (funClus != "pam") |
|
| 151 |
+# centrotypes <- c(centrotypes,indC) |
|
| 152 |
+# internalSim <- sim[indC,indC] |
|
| 153 |
+# } |
|
| 154 |
+# externalSim <- mean(sim[indC,setdiff(1:ncol(sim),indC)]) |
|
| 155 |
+# iq <- internalSim-externalSim |
|
| 156 |
+# |
|
| 157 |
+# return(iq) |
|
| 158 |
+# } |
|
| 126 | 159 |
|
| 127 | 160 |
|
| 128 |
- ## extract W including the centrotypes of each cluster |
|
| 161 |
+ ## Extract W including the centrotypes of each cluster |
|
| 129 | 162 |
W <- whit %*% allW[,centrotypes] |
| 130 | 163 |
A <- solve(t(W)%*%W) %*% t(W) |
| 131 | 164 |
S <- X%*%W |
