clusterTimeSeries: Cluster time series features.

In nlhuong/vistimeseq: Analysis for short time-series data

Description

Find the cluster assignment for timecourse features. Clustering computed for top "n.top.feat" features most variable over time in each of the selected "groups" using time-series expression (collpased over replicates). The cluster assignment of the remaining genes is based on the distance to the closest cluster centroid previously obtained. Hierarchical clustering is performed and both static and dynamic branch cutting algorithm are available for assigning cluster membership.

Usage

clusterTimeSeries(
  object,
  n.top.feat = 1000,
  groups.selected = "all",
  lambda = c(0.5, 0.25),
  clust.params = list()
)

Arguments

object	A TimeSeriesExperiment object
n.top.feat	A number of top most variable time-course features to use for clustering.
groups.selected	One or multiple groups from object@group to take into account when aggregating time-course features.
lambda	Weights for each lag difference, for time-course data. Length of lambda specifies number of lags to include. By default lag of order one and two are included with coefficients 0.5 and 0.25 respectively.
clust.params	A list contating arguments for hierarchical clustering. For details see clusterData.

Value

a TimeSeriesExperiment object with cluster assignment stored in cluster.map slot.

Examples

data("endoderm_small")
endoderm_small <- clusterTimeSeries(endoderm_small)
head(clusterMap(endoderm_small))

nlhuong/vistimeseq documentation built on Sept. 4, 2021, 2:41 a.m.