metaInteractions: meta-analysis for regulatory region and gene interactions

In BIMSBbioinfo/reg2gene: Predicting the target genes for regulatory elements

Description

The function combines association P-values and coefficients from different data sets using Fisher's method and weigthed averaging respectively. It is useful to combine datasets produced by different research groups while avoiding problems such as batch effects. For example, it can be used to combine Roadmap and Blueprint associateReg2Gene gene~enhancer single-model information obtained using across cell RNA-Seq signals and CHiP-Seq H3K27ac tracks. Aggregating single-model information (gene-enhancer pairs) by means of meta-analysis across different data-sources should increase the statistical power, improve the precision and accuracy of estimates and altogether produce more robust and reproducible results. After,combining P-values, q-values are calculated using the qvalue function.

Usage

metaInteractions(interactions)

Arguments

interactions

A list of GInteractions objects outputed from associateReg2Gene function.

Value

returns a GInteractions object with an updated statistics from the meta-analysis. The output will be similar to the output of associateReg2Gene function, it will have the same meta-data columns in the same order.

Author(s)

Altunislav Akalinski

Examples

# creating datasets

require(GenomicRanges)
require(InteractionSet)

# INPUT 0: 2 GRanges objects with toy expression values (x,y,z,a)

gr2 <- gr <- GRanges(seqnames=rep("chr1",3),IRanges(1:3,3:5))
   x <- 1:5
   y <- 2:6
   z <- 10:14
   a <- rep(0,length(x))
   
   
   GeneInfo <- as.data.frame(matrix(c(rep("gene",3),rep("regulatory",6)),
               ncol = 3,byrow = TRUE),stringsAsFactors=FALSE)
               colnames(GeneInfo) <- c("featureType","name","name2")
 
 mcols(gr) <- DataFrame(cbind(GeneInfo,rbind(x,y,z)))
 mcols(gr2) <- DataFrame(cbind(GeneInfo,rbind(x,y,a)))
 
# RUNNING associateReg2Gene and obtaining output results
 
 AssocObject <- reg2gene::associateReg2Gene(gr)
 AssocObject2 <- reg2gene::associateReg2Gene(gr2)
 
 # input for meta-analysis is list of such objects
 
 interactions <- list(AssocObject,AssocObject2)
 names(interactions) <- c("AssocObject","AssocObject2")
 
 # OUTPUT: Run metaA
 
 metaInteractions(interactions)

BIMSBbioinfo/reg2gene documentation built on May 3, 2019, 6:42 p.m.