R/other_cluster_based_subtyping.R
In rkawalerski/pdacR: Obtains and analyzes PDAC transcriptomics data
Defines functions
plot.hist
visualize.it
other_cluster_based_subtyping
Documented in
other_cluster_based_subtyping
#' @title Cluster Based Subtyping
#'
#' @description Perform clustering, label samples, and re-save data sets
#' @export
other_cluster_based_subtyping <- function(data_set_list = c("summary")){
pdf("other_cluster_based_subtyping.pdf")
#######################################################################
dataset <- pdac::Chen_GEO_array
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"Chen_GEO_array",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix((dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$specimen_type %in% "primaryPDAC" )
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("specimen_type is primaryPDAC",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[3]]$consensusTree
cluster.cut <- data.frame(cuts = (c("basal","classical",NA)[cutree(cluster.result.c,k=4)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('basal','classical')[cutree(cluster.result.c,k=7)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
Chen_GEO_array_plus <- dataset
save(file = "data/Chen_GEO_array_plus.RData",list = "Chen_GEO_array_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
#######################################################################
dataset <- pdac::Moffitt_S2
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"Moffitt_S2",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix(log2(1+dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$sample_type %in% "PDX")
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
sampleset <- which(dataset$sampInfo$sample_type %in% "CAF")
dataset$sampInfo$tumor.classifier.outlier[sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("sample_type is PDX",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('classical',NA,'basal','basal')[cutree(cluster.result.c,k=4)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('classical','basal')[cutree(cluster.result.c,k=3)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
Moffitt_S2_plus <- dataset
save(file = "data/Moffitt_S2_plus.RData",list = "Moffitt_S2_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
#######################################################################
dataset <- pdac::Moffitt_GEO_array
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"Moffitt_GEO_array",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix((dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$specimen_type %in% "Primary")
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
sampleset <- which(dataset$sampInfo$specimen_type %in% "Normal")
dataset$sampInfo$tumor.classifier.outlier[sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("specimen_type is Primary",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('classical','basal',NA)[cutree(cluster.result.c,k=7)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('basal','classical')[cutree(cluster.result.c,k=8)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
Moffitt_GEO_array_plus <- dataset
save(file = "data/Moffitt_GEO_array_plus.RData",list = "Moffitt_GEO_array_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
#######################################################################
dataset <- pdac::TCGA_PAAD
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"TCGA_PAAD",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix(log2(1+dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$Decision %in% "whitelist")
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
dataset$sampInfo$tumor.classifier.outlier[-sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("Decision is whitelist",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c(NA,'classical','basal')[cutree(cluster.result.c,k=7)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('basal','classical')[cutree(cluster.result.c,k=10)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
TCGA_PAAD_plus <- dataset
save(file = "data/TCGA_PAAD_plus.RData",list = "TCGA_PAAD_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
############################################################################
dev.off()
}
visualize.it <- function(dataset,smallx,cluster.result.c,featureset,cluster.cut){
cluster.result.r <- hclust(d = bioDist::cor.dist(x = (smallx)))
heatmap.3(x = smallx, scale="row",labRow = dataset$featInfo$SYMBOL[featureset],
col = colorRampPalette(c("blue", "white", "red"))(n = 299),
Colv = as.dendrogram(cluster.result.c),
Rowv = as.dendrogram(cluster.result.r),
ColSideColors = getSideColors(sampInfo = data.frame(cuts =
factor(x = cluster.cut$cuts,
levels = c("basal","classical","activated","normal"))),
sampleTracks = "cuts",
colorlists = list(c("orange","blue","brown","skyblue")),
drop.levels = FALSE)$SideColors,
RowSideColors = t(getSideColors(sampInfo = data.frame(basal =dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Basal.25,
classical =dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Classical.25,
normal =dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Normal.25,
activated = dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Activated.25),
sampleTracks = c("basal",
"classical",
"normal",
"activated"),
colorlists = list(b=c("white","orange"),
c=c("white","blue"),
n=c("white","skyblue"),
a=c("white","brown")))$SideColors))
}
plot.hist <- function(dataset){
par(mfrow=c(2,1))
h <- hist(log2(1+as.matrix(dataset$ex)),breaks=100,plot=FALSE)
plot(y=sqrt(h$count),x=h$mids,type='h',main = "Logged")
h <- hist(as.matrix(dataset$ex),breaks=100,plot=FALSE)
plot(y=sqrt(h$count),x=h$mids,type='h',main = "As Is")
}
rkawalerski/pdacR documentation built on Jan. 25, 2025, 10:50 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions plot.hist visualize.it other_cluster_based_subtyping
Documented in other_cluster_based_subtyping
#' @title Cluster Based Subtyping
#'
#' @description Perform clustering, label samples, and re-save data sets
#' @export
other_cluster_based_subtyping <- function(data_set_list = c("summary")){
pdf("other_cluster_based_subtyping.pdf")
#######################################################################
dataset <- pdac::Chen_GEO_array
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"Chen_GEO_array",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix((dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$specimen_type %in% "primaryPDAC" )
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("specimen_type is primaryPDAC",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[3]]$consensusTree
cluster.cut <- data.frame(cuts = (c("basal","classical",NA)[cutree(cluster.result.c,k=4)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('basal','classical')[cutree(cluster.result.c,k=7)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
Chen_GEO_array_plus <- dataset
save(file = "data/Chen_GEO_array_plus.RData",list = "Chen_GEO_array_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
#######################################################################
dataset <- pdac::Moffitt_S2
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"Moffitt_S2",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix(log2(1+dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$sample_type %in% "PDX")
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
sampleset <- which(dataset$sampInfo$sample_type %in% "CAF")
dataset$sampInfo$tumor.classifier.outlier[sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("sample_type is PDX",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('classical',NA,'basal','basal')[cutree(cluster.result.c,k=4)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('classical','basal')[cutree(cluster.result.c,k=3)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
Moffitt_S2_plus <- dataset
save(file = "data/Moffitt_S2_plus.RData",list = "Moffitt_S2_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
#######################################################################
dataset <- pdac::Moffitt_GEO_array
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"Moffitt_GEO_array",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix((dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$specimen_type %in% "Primary")
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
sampleset <- which(dataset$sampInfo$specimen_type %in% "Normal")
dataset$sampInfo$tumor.classifier.outlier[sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("specimen_type is Primary",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('classical','basal',NA)[cutree(cluster.result.c,k=7)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('basal','classical')[cutree(cluster.result.c,k=8)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
Moffitt_GEO_array_plus <- dataset
save(file = "data/Moffitt_GEO_array_plus.RData",list = "Moffitt_GEO_array_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
#######################################################################
dataset <- pdac::TCGA_PAAD
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,9,"TCGA_PAAD",adj = 0,cex = 3)
p <- p;text(1,5,paste("Total Samples",length(dataset$sampInfo[[1]])),adj = 0)
# -------------------------------------------------------------
# Boilerplate
plot.hist(dataset)
normalized.expression <- preprocessCore::normalize.quantiles(as.matrix(log2(1+dataset$ex)))
dataset$sampInfo$tumor.classifier.training <- FALSE
dataset$sampInfo$tumor.classifier.outlier <- FALSE
dataset$sampInfo[["cluster.MT"]] <- as.character(NA)
dataset$sampInfo[["cluster.MT.scaled"]] <- as.character(NA)
# -------------------------------------------------------------
# Select cases to cluster based on metadata
summary(dataset$sampInfo)
sampleset <- which(dataset$sampInfo$Decision %in% "whitelist")
dataset$sampInfo$tumor.classifier.training[sampleset] <- TRUE
dataset$sampInfo$tumor.classifier.outlier[-sampleset] <- TRUE
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("Decision is whitelist",
sum(dataset$sampInfo$tumor.classifier.training)),adj = 0)
# -------------------------------------------------------------
# Boilerplate
sampleset <- which(dataset$sampInfo$tumor.classifier.training)
featureset <- which(dataset$featInfo$SYMBOL %in%
c(as.character(pdac::gene_lists$Moffitt.Basal.25),
as.character(pdac::gene_lists$Moffitt.Classical.25)))
smallx <- (normalized.expression[featureset,sampleset])
smallx.scaled <- t(scale(t(normalized.expression[featureset,sampleset])))
# -------------------------------------------------------------
# unscaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c(NA,'classical','basal')[cutree(cluster.result.c,k=7)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# scaled clustering
cluster.result.c <- ConsensusClusterPlus::ConsensusClusterPlus(d = as.matrix(smallx.scaled),
seed = 1234,
pFeature = 0.8,
pItem = 0.8,
maxK = 3,
reps=200,
distance="pearson",
clusterAlg="kmdist")[[2]]$consensusTree
cluster.cut <- data.frame(cuts = (c('basal','classical')[cutree(cluster.result.c,k=10)]),stringsAsFactors = FALSE)
dataset$sampInfo$cluster.MT.scaled[sampleset] <- cluster.cut[[1]]
visualize.it(dataset,smallx,cluster.result.c,featureset,cluster.cut)
# -------------------------------------------------------------
# adjustments to training set
dataset$sampInfo$tumor.classifier.outlier[is.na(dataset$sampInfo$cluster.MT) &
is.na(dataset$sampInfo$cluster.MT.scaled) &
dataset$sampInfo$tumor.classifier.training ] <- TRUE
dataset$sampInfo$tumor.classifier.training[is.na(dataset$sampInfo$cluster.MT)] <- FALSE
# -------------------------------------------------------------
# Wrapup
dataset$sampInfo$tumor.classifier.training <- factor(dataset$sampInfo$tumor.classifier.training,
levels = c(FALSE,TRUE))
dataset$sampInfo$tumor.classifier.outlier <- factor(dataset$sampInfo$tumor.classifier.outlier,
levels = c(FALSE,TRUE))
dataset$sampInfo$cluster.MT <- factor(dataset$sampInfo$cluster.MT, levels = c("basal","classical"))
dataset$sampInfo$cluster.MT.scaled <- factor(dataset$sampInfo$cluster.MT.scaled, levels = c("basal","classical"))
TCGA_PAAD_plus <- dataset
save(file = "data/TCGA_PAAD_plus.RData",list = "TCGA_PAAD_plus")
p <- plot(x=1:10,y=1:10,ann=FALSE,type="n",xaxt="n",yaxt="n")
p <- p;text(1,8,paste("after cluster trimming",sum(!is.na(dataset$sampInfo$cluster.MT))),adj = 0)
############################################################################
dev.off()
}
visualize.it <- function(dataset,smallx,cluster.result.c,featureset,cluster.cut){
cluster.result.r <- hclust(d = bioDist::cor.dist(x = (smallx)))
heatmap.3(x = smallx, scale="row",labRow = dataset$featInfo$SYMBOL[featureset],
col = colorRampPalette(c("blue", "white", "red"))(n = 299),
Colv = as.dendrogram(cluster.result.c),
Rowv = as.dendrogram(cluster.result.r),
ColSideColors = getSideColors(sampInfo = data.frame(cuts =
factor(x = cluster.cut$cuts,
levels = c("basal","classical","activated","normal"))),
sampleTracks = "cuts",
colorlists = list(c("orange","blue","brown","skyblue")),
drop.levels = FALSE)$SideColors,
RowSideColors = t(getSideColors(sampInfo = data.frame(basal =dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Basal.25,
classical =dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Classical.25,
normal =dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Normal.25,
activated = dataset$featInfo$SYMBOL[featureset] %in%
pdac::gene_lists$Moffitt.Activated.25),
sampleTracks = c("basal",
"classical",
"normal",
"activated"),
colorlists = list(b=c("white","orange"),
c=c("white","blue"),
n=c("white","skyblue"),
a=c("white","brown")))$SideColors))
}
plot.hist <- function(dataset){
par(mfrow=c(2,1))
h <- hist(log2(1+as.matrix(dataset$ex)),breaks=100,plot=FALSE)
plot(y=sqrt(h$count),x=h$mids,type='h',main = "Logged")
h <- hist(as.matrix(dataset$ex),breaks=100,plot=FALSE)
plot(y=sqrt(h$count),x=h$mids,type='h',main = "As Is")
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.