CytoK: CytoK
In Ghoshlab/cytoKernel: Differential expression using kernel-based score test

Description Usage Arguments Details Value References Examples

This function applies a kernel-based score test for identifying differentially expressed features in high-throughput experiments, called the the CytoK procedure. This function also defines the CytoK class and constructor.

CytoK(
  object,
  group_factor,
  lowerRho = 2,
  upperRho = 12,
  gridRho = 4,
  alpha = 0.05,
  featureVars = NULL
)

`object`	an object which is a `matrix` or `data.frame` with features (e.g. cluster-marker combinations or genes) on the rows and samples as the columns. Alternatively, a user can provide a `SummarizedExperiment` object and the `assay(object)` will be used as input for the CytoK procedure.
`group_factor`	a group level binary categorical response associated with each sample or column in the `object`. The order of the `group_factor` must match the order of the columns in `object`.
`lowerRho`	(Optional) lower bound of the kernel parameter.
`upperRho`	(Optional) upper bound of the kernel parameter.
`gridRho`	(Optional) number of grid points in the interval [lowerRho, upperRho].
`alpha`	(Optional) level of significance to control the False Discovery Rate (FDR). Default is 0.05.
`featureVars`	(Optional) Vector of the columns which identify features. If a 'SummarizedExperiment' is used for 'data', row variables will be used.

CytoK (Kernel-based score test in biological feature differential analysis) is a nonlinear approach, which identifies differentially expressed features in high-dimensional biological experiments. This approach can be applied across many different high-dimensional biological data including Flow/Mass Cytometry data and other variety of gene expression data. The CytoK procedure employs a kernel-based score test to identify differentially expressed features. This procedure can be easily applied to a variety of measurement types since it uses a Gaussian distance based kernel.

This function computes the feature-wise p values and their corresponding adjusted p values. Additionally, it also computes the feature-wise shrunk effect sizes and their corresponding shrunk effect size sd's. Further, it calculates the percent of differentially expressed features. See the vignette for more details.

A object of the class CytoK that contains a data.frame of the CytoK features in the CytoKFeatures slot, a data.frame of the CytoK features in the CytoKFeaturesOrdered slot ordered by adjusted p values from low to high, a numeric value of the CytoK differentially expressed features CytoKDEfeatures slot, a data.frame or SummarizedExperiment original data objject in the CytoKData slot, a numeric value of the level of significance in the CytoKalpha slot and (optional) a vector of the columns which identify features in the CytoKfeatureVars slot.

Liu D, Ghosh D, Lin X. Estimation and testing for the effect of a genetic pathway on a disease outcome using logistic kernel machine regression via logistic mixed models. BMC Bioinf. 2008; 9(1):292.

Zhan X, Ghosh D. Incorporating auxiliary information for improved prediction using combination of kernel machines. Stat Methodol. 2015; 22:47–57.

Zhan, X., Patterson, A.D. & Ghosh, D. Kernel approaches for differential expression analysis of mass spectrometry-based metabolomics data. BMC Bioinformatics 16, 77 (2015). https://doi.org/10.1186/s12859-015-0506-3

Matthew Stephens, False discovery rates: a new deal, Biostatistics, Volume 18, Issue 2, April 2017, Pages 275–294, https://doi.org/10.1093/biostatistics/kxw041

data <- cbind(matrix(rnorm(1200,mean=2, sd=1.5),
nrow=200, ncol=6), matrix(rnorm(1200,mean=5, sd=1.9),
nrow=200, ncol=6))
data_CytoK <- CytoK(object=data,
group_factor = rep(c(0,1), each=6), lowerRho=2,
upperRho=12,gridRho=4,alpha = 0.05,
featureVars = NULL)
data("cytoHDBMW")
data_CytoK_HD <- CytoK(object=cytoHDBMW,
group_factor = rep(c(0, 1), c(4, 4)), lowerRho=2,
upperRho=12,gridRho=4,alpha = 0.05,
featureVars = NULL)