citrus.predict: Predict labels of new feature set

In nolanlab/citrus: Citrus: Identification of stratifying subpopulations in Flow Cytometry Data.

Predict labels of new feature set

Description

Predict labels of new feature set

Usage

citrus.predict.classification(citrus.endpointModel, newFeatures)

citrus.predict.continuous(citrus.endpointModel, newFeatures)

citrus.predict(citrus.endpointModel, newFeatures)

Arguments

citrus.endpointModel	A citrus.endpointModel object.
newFeatures	Features from samples to predict labels for.

Value

Matrix of predicted sample endpoints at all model regularization thresholds.

Author(s)

Robert Bruggner

Examples

# Where the data lives
dataDirectory = file.path(system.file(package = "citrus"),"extdata","example1")

# List of files to be clustered
fileList1 = data.frame("unstim"=list.files(dataDirectory,pattern=".fcs")[seq(from=2,to=20,by=2)])

# List of files to be mapped
fileList2 = data.frame("unstim"=list.files(dataDirectory,pattern=".fcs")[seq(from=1,to=19,by=2)])

# Read the data
citrus.combinedFCSSet1 = citrus.readFCSSet(dataDirectory,fileList1)
citrus.combinedFCSSet2 = citrus.readFCSSet(dataDirectory,fileList2)

# List of columns to be used for clustering
clusteringColumns = c("Red","Blue")

# Cluster first dataset
citrus.clustering = citrus.cluster(citrus.combinedFCSSet1,clusteringColumns)

# Map new data to exsting clustering
citrus.mapping = citrus.mapToClusterSpace(citrus.combinedFCSSet.new=citrus.combinedFCSSet2,citrus.combinedFCSSet.old=citrus.combinedFCSSet1,citrus.clustering)

# Large Enough Clusters
largeEnoughClusters = citrus.selectClusters(citrus.clustering)

# Clustered Features and mapped features
clusteredFeatures = citrus.calculateFeatures(citrus.combinedFCSSet1,clusterAssignments=citrus.clustering$clusterMembership,clusterIds=largeEnoughClusters)
mappedFeatures = citrus.calculateFeatures(citrus.combinedFCSSet2,clusterAssignments=citrus.mapping$clusterMembership,clusterIds=largeEnoughClusters)

# Labels
labels = factor(rep(c("Healthy","Diseased"),each=10))

# Build Endpoint Model
citrus.endpointModel = citrus.buildEndpointModel(clusteredFeatures,labels[seq(from=2,to=20,by=2)])

# Predict
citrus.predict(citrus.endpointModel,newFeatures=mappedFeatures)

nolanlab/citrus documentation built on April 30, 2022, 3:24 a.m.