R/negbin.r

In carjed/smaug: Helper functions for the SMAUG pipeline

##############################################################################
# Negbin model processing functions
##############################################################################
# Run negbin model for each formula specified in formlist
runMod <- function(formlist, data){
	out <- lapply(formlist, function(x) glm.nb(x, data))
	names(out) <- names(formlist)
	return(out)
}

# Build list of fitted values (with CHR/BIN names) for each model
getFits <- function(modlist, data){
	out <- lapply(modlist,
		function(x){
			y <- fitted.values(x)
			names(y) <- paste0(data$CHR, ".", data$BIN)
			y
		})
	names(out) <- names(modlist)
	return(out)
}

# Build list of dataframes for each model
buildDF <- function(fitlist, data){
	out <- lapply(fitlist,
		function(x){
			data.frame(CHR, Category2=cat1, BIN,
				exp=x,
				obs=data$obs,
				# res=names(fitlist),
				stringsAsFactors=F)
		}
	)
	names(out) <- names(fitlist)
	return(out)
}

##############################################################################
# Compute standard error for correlations
##############################################################################
corSE <- function(corval, ct){
	sqrt((1-corval^2)/(ct-2))
}