makeClusters {clusterSeq} | R Documentation |
This function uses minimal linkage data to perform rapid clustering by
singleton agglomeration (i.e., a gene will always cluster with its
nearest neighbours provided the distance to those neighbours does not
exceed some threshold). For alternative (but slower) clustering options,
see the makeClustersFF
function.
makeClusters(aM, cD, threshold = 0.5)
aM |
A data frame constructed by |
cD |
The data given as input to |
threshold |
A threshold on the maximum dissimilarity at which two genes can cluster. Defaults to 0.5. |
An IntegerList object, each member of whom defines a cluster of co-expressed genes. The object is ordered decreasingly by the size of each cluster.
Thomas J Hardcastle
makeClustersFF
kCluster
associatePosteriors
#Load in the processed data of observed read counts at each gene for each sample. data(ratThymus, package = "clusterSeq") # Library scaling factors are acquired here using the getLibsizes # function from the baySeq package. libsizes <- getLibsizes(data = ratThymus) # Adjust the data to remove zeros and rescale by the library scaling # factors. Convert to log scale. ratThymus[ratThymus == 0] <- 1 normRT <- log2(t(t(ratThymus / libsizes)) * mean(libsizes)) # run kCluster on reduced set. normRT <- normRT[1:1000,] kClust <- kCluster(normRT) # make the clusters from these data. mkClust <- makeClusters(kClust, normRT, threshold = 1)