R/Seurat_FindAllMarkers.R

In stela2502/BioData: The package allows to annotate a numeric data.table with col and row data

#' @name Seurat_FindAllMarkers
#' @aliases Seurat_FindAllMarkers,BioData-method
#' @rdname Seurat_FindAllMarkers-methods
#' @docType methods
#' @description  Use Seurat's FindAllMarkers function
#' @param x the BioData object
#' @param condition The grouping you wat to analyze
#' @param logfc.threshold test only genes with a liog fold change of default=1.0 or more 
#' @param minPct test only genes with expression in more than default=0.1 fraction of cells in at least one group
#' @title description of function Seurat_FindAllMarkers
#' @export 
setGeneric('Seurat_FindAllMarkers', ## Name
	function ( x , condition, logfc.threshold=1.0, minPct=0.1) { 
		standardGeneric('Seurat_FindAllMarkers')
	}
)

setMethod('Seurat_FindAllMarkers', signature = c ('BioData'),
	definition = function ( x , condition, logfc.threshold = 1, minPct=0.1 ) {
	
	stop( "Please use the FAstWilcoxTest based test 10x speed improvement - same results!")
	
	# object <- as_Seurat( x, group=condition, fromRaw=F ) ## I want to use MY data
	
	# stats = Seurat::FindAllMarkers( object, logfc.threshold = logfc.threshold, minPct = minPct )

	# CppStats <- function( n ) {
	# 	print ( paste("Calc wilcox test for",n) )
	# 	OK = which(grp.vec == n )
	# 	BAD= which(grp.vec != n )
	# 	if ( length(OK) > 0 && length(BAD) > 0 ){
	# 		r = as.data.frame(
	# 				FastWilcoxTest::StatTest( Matrix::t( x$dat), OK, BAD, logfc.threshold, minPct )
	# 		)
	# 		r= r[order( r[,'p.value']),]
	# 		r = cbind( r, cluster= rep(n,nrow(r) ), gene=rownames(x$dat)[r[,1]] )
	# 	}
	# 	r
	# }
	
	# stats = NULL;
	# grp.vec = as.vector(x$samples[,condition])
	
	# for ( n in  unique( sort(grp.vec)) ) {
	# 	stats = rbind( stats, CppStats(n) )
	# }
	
	# x$stats[[name]] = stats
	
	invisible(x)
} )

#' @name Cpp_FindAllMarkers
#' @aliases Cpp_FindAllMarkers,BioData-method
#' @rdname Cpp_FindAllMarkers-methods
#' @docType methods
#' @description  Use Seurat's FindAllMarkers function
#' @param x the BioData object
#' @param condition The grouping you wat to analyze
#' @param logfc.threshold seurat option used for speed up (default = 1)
#' @param  minPct only test genes that are detected in a minimum fraction of
#' min.pct cells in either of the two populations. Meant to speed up the function
#' @param onlyPos check only higher expression (default FALSE)	
#' @title description of function Cpp_FindAllMarkers
#' @export 
setGeneric('Cpp_FindAllMarkers', ## Name
		function ( x , condition, logfc.threshold = 1, minPct = 0.1, onlyPos = FALSE) { 
			standardGeneric('Cpp_FindAllMarkers')
		}
)

setMethod('Cpp_FindAllMarkers', signature = c ('BioData'),
		definition = function ( x , condition, logfc.threshold = 1, minPct = 0.1 , onlyPos = FALSE) {
			
		CppStats <- function( n ) {
        	print ( paste("Calc wilcox test for",n) )
        	OK = which(grp.vec == n )
        	BAD= which(grp.vec != n )
            if ( length(OK) > 0 && length(BAD) > 0 ){
				r = NULL
                r = as.data.frame(
                     FastWilcoxTest::StatTest( Matrix::t( tmpM ), OK, BAD, logfc.threshold, minPct )
                )
				t= ''
				try ( { t = class(r)} ,silent=T )
				if ( ! t == 'data.frame'){
					browser()
				}
                r= r[order( r[,'p.value']),]
                r = cbind( r, cluster= rep(n,nrow(r) ), gene=rownames(x$dat)[r[,1]] )
			}
            r
        }

        stats = NULL;
        grp.vec = as.vector(x$samples[,condition])
		tmpM = x$dat
		tmpM@x[which(tmpM@x < 0)] = 0
        for ( n in  unique( sort(grp.vec)) ) {
                stats = rbind( stats, CppStats(n) )
        }


			stats = cbind( stats, 'p_val_adj' = stats::p.adjust( stats[,'p.value'], method='BH' ) )
			name = paste(sep="_", "Cpp_FindAllMarkers", condition) 
			x$stats[[name]] = stats
			
			invisible(x)
		} )