update2.4.md

In epurdom/clusterCells: Compare Clusterings for Single-Cell Sequencing

New to Version 2.3.0

Change to ClusterExperiment Object

Important If you have objects created since 2.0.0 but with a version < 2.3.0 (i.e. including 2.0.0), you should run updateObject to update the class definition because there have been changes to the class definition since that version:

ceObj<-updateObject(ceObj)

Warning This command will, however, loose information saved about the last mergeClusters call that you have made if your object is from version < 2.1.4. You may want to save that information and manually update the slots. If you do so, make sure you call validObject to make sure that you have done so correctly (in particular, you will have to have a value for the slot merge_demethod, see ?ClusterExperiment which is a new slot). For example,

ceObjNew<-updateObject(ceObj)
ceObjNew@merge_index<-ceObj@merge_index
<etc>

If you have objects from before 2.0.0 (when the class was called 'clusterExperiment'), you should construct a new object using the ClusterExperiment function. For example,

ceObjNew<-ClusterExperiment(
    se=as(ceObj,"SingleCellExperiment"), 
    clusterMatrix=ceObj@clusterMatrix, 
    <etc>
    )

See ?ClusterExperiment for the names of the slots.

Other important changes

There have also been a number of changes and enhancements to the package. These are the most important (a complete list is detailed in the NEWS file of the package -- all releases since May 1, 2018)

• We have changed the function combineMany to makeConsensus. This has resulted in changes to the names of the arguments of RSEC
  • combineProportion -> consensusProportion in RSEC
  • combineMinSize -> consensusMinSize in RSEC
• Add functionality to getBestFeatures to allow edgeR for DE, as well as weights used with edgeR for compatability with weights to handle zero-inflation. As part of this change isCount argument has been replaced with more fine-grained DEMethod argument in getBestFeatures, mergeClusters; and the argument mergeDEMethod in RSEC is now available.
• We have changed the argument sampleData in various plotting commands to colData to better indicate that the argument is to identify columns in colData that should also be plotted. Furthermore plotDendrogram now takes the argument colData for plotting of information in colData with the dendrogram.
• We have changed the names of arguments related to unassigned (-1 or -2 assignments) to more consistently use the term "unassigned", as well as adding the function assignUnassigned:
  • argument removeNegative -> removeUnassigned in getBestFeatures
  • argument ignoreUnassignedVar -> filterIgnoresUnassigned in mergeClusters (and other functions) for clarity.
  • function removeUnclustered -> removeUnassigned
• New plotting functions:
  • plotTableClusters
  • plotFeatureScatter
• Allow the arguments subsample and sequential to RSEC to allow for opting out of those options for large datasets (but default is TRUE unlike clusterMany)
• The argument whichAssay is added to most functions to allow the user to select the assay on which the operations will be performed.
• We've changed how we store the cluster hierarchies so that we now use the phylo4d class of phylobase package (previously we stored them as a dendrogram class). This makes it easier to store information about the dendrograms and manipulate them. There are various helper functions related to this change. See ?clusterDendrogram.
• We now store the coClustering matrix in the coClustering slot as a sparseMatrix class from the package Matrix. This will reduce the size of the object in memory.

epurdom/clusterCells documentation built on April 28, 2024, 8:14 p.m.