vamForSeurat {VAM}

VAM wrapper for scRNA-seq data processed using the Seurat framework

Description

Executes the Variance-adjusted Mahalanobis (VAM) method (vamForCollection) on normalized scRNA-seq data stored in a Seurat object. If the Seurat NormalizeData method was used for normalization, the technical variance of each gene is computed as the proportion of technical variance (from FindVariableFeatures) multiplied by the variance of the normalized counts. If SCTransform was used for normalization, the technical variance for each gene is set to 1 (the normalized counts output by SCTransform should have variance 1 if there is only technical variation).

Usage

    vamForSeurat(seurat.data, gene.weights, gene.set.collection, 
    	center=FALSE, gamma=TRUE, sample.cov=FALSE, return.dist=FALSE)

Arguments

seurat.data

The Seurat object that holds the scRNA-seq data. Assumes normalization has already been performed.

gene.weights

See description in vamForCollection

gene.set.collection

List of m gene sets for which scores are computed. Each element in the list corresponds to a gene set and the list element is a vector of indices for the genes in the set. The index value is defined relative to the order of genes in the relevant seurat.data Assay object. Gene set names should be specified as list names.

center

See description in vam

gamma

See description in vam

sample.cov

If true, will use the a diagonal covariance matrix generated from the sample variances to compute the squared adjusted Mahalanobis distances (this is equivalent to not specifying tech.var for the vam method). If false (default), will use the technical variances as determined based on the type of Seurat normalization.

return.dist

If true, will return the squared adjusted Mahalanobis distances in a new Assay object called "VAM.dist". Default is F.

Value

Updated Seurat object that hold the VAM results in one or two new Assay objects:

• If return.dist is true, the matrix of squared adjusted Mahalanobis distances will be stored in new Assay object called "VAM.dist".

• The matrix of CDF values (1 minus the one-sided p-values) will be stored in new Assay object called "VAM.cdf".

See Also

vam,vamForCollection

Examples

    # Only run example code if Seurat package is available
    if (requireNamespace("Seurat", quietly=TRUE) & requireNamespace("SeuratObject", quietly=TRUE)) {
        # Define a collection with one gene set for the first 10 genes
        collection=list(set1=1:10)
        # Execute on the pbmc_small scRNA-seq data set included with SeuratObject
        # See vignettes for more detailed Seurat examples
        vamForSeurat(seurat.data=SeuratObject::pbmc_small,
            gene.set.collection=collection)
    }