cisAssoc: test for variant-expression associations in cis, using VCF

Description

test for variant-expression associations in cis, using VCF and RangedSummarizedExperiment representations

Usage

cisAssoc(summex, vcf.tf, rhs = ~1, nperm = 3, cisradius =
                 50000, genome = "hg19", assayind = 1, lbmaf = 1e-06,
                 lbgtf = 1e-06, dropUnivHet = TRUE, infoFields =
                 c("LDAF", "SVTYPE"), simpleSNV = TRUE)
cisEsts(summex, vcf.tf, rhs = ~1, nperm = 3, cisradius =
                 50000, genome = "hg19", assayind = 1, lbmaf = 1e-06,
                 lbgtf = 1e-06, dropUnivHet = TRUE, infoFields =
                 c("LDAF", "SVTYPE"), simpleSNV = TRUE)
cisCount(summex, vcf.tf, rhs = ~1, cisradius =
                 50000, genome = "hg19", assayind = 1, lbmaf = 1e-06,
                 lbgtf = 1e-06, dropUnivHet = TRUE, infoFields =
                 c("LDAF", "SVTYPE"), simpleSNV = TRUE)
AllAssoc(summex, vcf.tf, variantRange, rhs = ~1, nperm = 3, 
    genome = "hg19", assayind = 1, lbmaf = 1e-06, lbgtf = 1e-06, 
    dropUnivHet = TRUE, infoFields = c("LDAF", "SVTYPE"))

Arguments

summex

a RangedSummarizedExperiment object

vcf.tf

instance of TabixFile, referring to a tabix-indexed, bgzipped VCF file

rhs

formula `right hand side' for adjustments to be made as snp.rhs.tests is run on each expression vector

nperm

number of permutations to be used for plug-in FDR computation

cisradius

distance in bp around each gene body to be searched for SNP association

genome

tag suitable for use in GenomeInfoDb structures

assayind

index of assays(summex) to use for expression data retrieval

lbmaf

lower bound on MAF of SNP to retain for analysis, computed using col.summary

lbgtf

lower bound on genotype frequency of SNP to retain for analysis

dropUnivHet

logical, if TRUE, will check for columns of SnpMatrix instance that possess no values other than "NA" and "A/B". See http://www.biostars.org/p/117155/#117270

infoFields

character -- VCF fields to retain in vcfInfo() part of query

simpleSNV

logical -- will use simple computation of isSNV to filter variants for analysis to SNV

variantRange

GRanges instance that defines the scope of the VCF to be used for testing against all features on summex

Value

cisAssoc: a GRanges-class instance with mcols including chisq, permScore...
cisCount: enumerate locations in VCF that would be tested

Details

snp.rhs.tests is the workhorse for statistical modeling. VCF content is transformed to the byte-code (which allows for uncertain imputation) and used in fast testing.

Examples

Run this code

require(GenomeInfoDb)
 require(geuvPack)
 require(Rsamtools)
 data(geuFPKM)
 lgeu = geuFPKM[ which(seqnames(geuFPKM)=="chr20"), ]
 seqlevelsStyle(lgeu) = "NCBI"
 tf20 = TabixFile(system.file("vcf/c20exch.vcf.gz", package="gQTLstats"))
 if (require(VariantAnnotation)) scanVcfHeader(tf20)
 lgeue = clipPCs(lgeu[,which(lgeu$popcode=="CEU")], 1:2)
 set.seed(1234)
 litc = cisAssoc(lgeue[c(162,201),], tf20, nperm=2, lbmaf=.05, cisradius=50000)
 set.seed(1234)
 lite = cisEsts(lgeue[c(162,201),], tf20, nperm=2, lbmaf=.05, cisradius=50000)
 summary(litc$chisq)
 mysr = range(litc)
 litc$pifdr = gQTLstats:::pifdr(litc$chisq, c(litc$permScore_1, litc$permScore_2))
 litc[which(litc$pifdr < .01)]
 lita = AllAssoc(geuFPKM[1:10,], tf20, mysr)
 lita3 = AllAssoc(geuFPKM[11:20,], tf20, mysr)
 #lita5 = AllAssoc(geuFPKM[21:30,], tf20, mysr)
 n1 = gQTLstats:::collapseToBuf(lita, lita3)
 #n1 = collapseToBuf(n1, lita5)