SMITE (version 1.0.2)

annotateModification: Adding modification data to a PvalueAnnotation

Description

This function is the main "workhorse" function for SMITE because given a specific epigenetic modification (e.g. DNA methylation) it will 1) assess an internal correlation structure and 2) aggregate the modification over all intervals associated with a gene in the "makePvalueAnnotation" function.

Usage

annotateModification(pvalue_annotation, mod_data, weight_by = NULL, 
weight_by_method = "Stouffer", mod_included = NULL, mod_corr = TRUE, 
mod_type = "methylation", verbose = FALSE)

Arguments

pvalue_annotation

An S4 object of class PvalueAnnotation.

mod_data

A dataframe or matrix derived from a bed file with thhe the first three columns as (chromosome, start, end), column 4 is the effect, and column 5 is the p-value.

weight_by

A vector with named elements specifying how modifications should be weighted within an interval. Must be one of: "distance" Use the distance from the gene TSS to weight the p-values and the combined effect such that events closer to the TSS are weighted more. Log distances are used. "pval" "p.value" "pvalue" "p_val" (DEFAULT) Do not weight p-values but weight the combined effect such by the signficance of that effect. ELSE Do not weight p-values or the combined effect.

weight_by_method

A character specifying which method should be used to combine p-values. Must be one of:

"Stouffer" (DEFAULT) Stouffer's method for combing pvalues involves first taking the inverse standard normal CDF transformation of a vector of p-values followed by a weighted sum creating a new Z score with a standard normal distribution "Fisher" "fisher" "chisq" "chi" Fisher's method involves summing the -2ln(p) for each of k p's which follows an approximate chi square distribution with 2k degrees of freedom "Sidak" "sidak" "minimum" Sidak's adjustment is essentially the minimum p-value, with an added transformation to account for multiple comparisons. "binomial" The binomial probability assesses the probability of observing the observed number of p-value below a threshold (alpha=0.05) given the total number of p values and the probability of a false positive.

mod_included

A vector of named elements specifying for which intervals in the annotation the function should find combined scores (e.g. promoters). If not specified the assumption is that all type of intervals associated with a gene should be included.

mod_corr

A logical (TRUE/FALSE) specifying whether a correlation matrix should be estimated. The DEFAULT is TRUE.

mod_type

A character naming the modification that is being loaded. The DEFAULT is "methylation" and any modType string can be used, but will be referred to in downstream analysis. A unique name must be used for each modification that is loaded. When picking a variable modType should also avoid using "_" as it is used to split column names containing modType.

verbose

A logical specifying if the user wants updates about the progress of the function.

Value

A an S4 object of class PvalueAnnotation with the slot modification (a GrangesList) filled in for each additional modification.

Details

This function is the main "workhorse" function for SMITE because given a specific epigenetic modification (e.g. DNA methylation) it will 1) assess an internal correlation structure and 2) aggregate the modification over all intervals associated with a gene in the "makePvalueAnnotation" function.

References

Fisher R. Statistical methods for research workers. Oliver and Boyd; Edinburgh: 1932.

Stouffer S, DeVinney L, Suchmen E. The American soldier: Adjustment during army life. Vol. 1. Princeton University Press; Princeton, US: 1949.

Sidak, Z. (1967). Rectangular confidence regions for the means of multivariate normal distributions, Journal of the American Statistical Association 62, 626 633.

See Also

removeModification annotateExpression makePvalueAnnotation createPvalueObject

Examples

options(stringsAsFactors=FALSE)

## Commented out below See vignette for more detailed usage information ##
## Load genome bed file ##
#data(hg19_genes_bed)

## Create a PvalueAnnotation with defaults for promoter size##
#test_annotation<-makePvalueAnnotation(data=hg19_genes, gene_name_col=5)

## Load DNA methylation bed file ##
#data(methylationdata)
#methylation<-methylation[-which(is.na(methylation[,5])),]
#methylation[,5]<-replace(methylation[,5],methylation[, 5] == 0, 
#min(subset(methylation[,5], methylation[,5]!=0), na.rm=TRUE))

## Load DNA methylation into PvalueAnnotation modCorr=F for example##
## NOTE: Commented out below.  See vignette for better example ##
#test_annotation <- annotateModification(pvalue_annotation=test_annotation, 
#mod_data=methylation, weight_by=c(promoter="distance", body="distance"),
#verbose=FALSE, mod_corr=FALSE, mod_type="methylation")