Gene-based Association Test that uses an extended Simes procedure (GATES) for single trait - SNP set association

Description

Get the p-value of GATES. Usually it is used to get genomewise p-values. This function is taken from postgwas package. There is a little modification of the code GATES in postgwas package. 1) The approximated matrix may have negative eigen value, we modified it not to have negative values; 2) we added one more return (the key gene location) for Hyst method.

Usage

GATES2(ldmatrix, p)

Arguments

Value

A p-value of GATES and the key gene location (to be used by Hyst).

Author(s)

Milan Hiersche(taken from pastgwas package), Il-Youp Kwak(modified a little)

References

Miao-Xin Li, Hong-Sheng Gui, Johnny S.H. Kwan and Pak C. Sham (2011) GATES: A Rapid and Powerful Gene-Based Association Test Using Extended Simes Procedure The American Journal of Human Genetics 88, 283-293

See Also

Hyst GatesSimes

Examples

#simula <- simPathAR1Snp(nGenes=20, nGenes1=1, nSNPlim=c(1, 20), nSNP0=1:3,
#                           LOR=.2, rholim=c(0,0),
#                           n=30, MAFlim=c(0.05, 0.4), p0=0.05)
#Ps <- getlogitp(simula$Y, simula$X)

## get correlation of SNPs using controls
#ldmat <- cor(simula$X[ simula$Y == 0, ])

#o.pvec = order(Ps)
# ldmat <- ldmat[o.pvec, o.pvec]
#(gatesp <- GATES2(ldmat, sort(Ps))[1])

The SPU and aSPU tests for multiple traits - single SNP association in generalized estimating equations.

Description

It gives p-values of the GEESPU tests and GEEaSPU test.

Usage

GEEaSPU(
  traits,
  geno,
  Z = NULL,
  model = c("binomial", "gaussian"),
  gamma = c(1:8, Inf),
  n.sim = 1000,
  corstr = "independence"
)

Arguments

Value

p-values for the GEE-SPU and GEE-aSPU test.

Author(s)

Junghi Kim, Wei Pan and Il-Youp Kwak

References

Yiwei Zhang, Zhiyuan Xu, Xiaotong Shen, Wei Pan (2014) Testing for association with multiple traits in generalized estimation equations, with application to neuroimaging data. Neuroimage. 96:309-25

Examples

traits <- matrix(rnorm(100*5, 0,1), ncol=5)
Z <- rnorm(100, 2, 0.5)
geno <- rbinom(100, 2, 0.5)
out <- GEEaSPU(traits, geno, Z = NULL, model = "gaussian", 
	  gamma = c(1:8,Inf), n.sim = 100)

GATES-Simes test for single trait - pathway association.

Description

Get the p-value of GATES-Simes. It uses an extended Simes procedure to combine GATES p-values across multiple genes in a pathway.

Usage

GatesSimes(pvec, ldmatrix, snp.info, gene.info)

Arguments

Value

A p-value.

Author(s)

Il-Youp Kwak and Wei Pan

References

Hongsheng Gui, Miaoxin Li, Pak C Sham and Stacey S Cherny (2011) Comparisons of seven algorithms for pathway analysis using the WTCCC Crohn's Disease BMC Research Notes, 4:386

See Also

Hyst GATES2

Examples

#simula <- simPathAR1Snp(nGenes=20, nGenes1=1, nSNPlim=c(1, 20), nSNP0=1:3,
#                           LOR=.2, rholim=c(0,0),
#                           n=30, MAFlim=c(0.05, 0.4), p0=0.05)
#logitp <- getlogitp(simula$Y, simula$X)

## get correlation of SNPs using controls
#ldmat <- cor(simula$X[ simula$Y == 0, ])
#out <- GatesSimes(pvec = logitp, ldmatrix = ldmat, snp.info = simula$snp.info,
#                  gene.info = simula$gene.info)
#out

HYST (Hybrid Set-based Test) for single trait - pathway association

Description

Get a p-value using HYST.

Usage

Hyst(pvec, ldmatrix, snp.info, gene.info)

Arguments

Value

A p-value.

Author(s)

Il-Youp Kwak and Wei Pan

References

Miao-Xin Li, Johnny S.H. Kwan and Pak C. Sham (2012) HYST: A Hybrid Set-Based Test for Genome-wide Association Studies, with Application to Protein-Protein Interaction-Based Association Analysis The American Journal of Human Genetics, 91, 478-488.

See Also

GatesSimes GATES2

Examples

#simula <- simPathAR1Snp(nGenes=20, nGenes1=1, nSNPlim=c(1, 20), nSNP0=1:3,
#                           LOR=.2, rholim=c(0,0),
#                           n=30, MAFlim=c(0.05, 0.4), p0=0.05)
#logitp <- getlogitp(simula$Y, simula$X)

## get correlation of SNPs using controls
#ldmat <- cor(simula$X[ simula$Y == 0, ])
#out <- Hyst(pvec = logitp, ldmatrix = ldmat, snp.info = simula$snp.info,
#            gene.info = simula$gene.info)

The SPU and aSPU tests for multiple traits - single SNP association with GWAS summary statistics.

Description

SNP based adaptive association test for multiple phenotypes with GWAS summary statistics.

Usage

MTaSPUs(Z, v, B, pow, transform = FALSE, Ps = FALSE)

Arguments

Value

compute p-values for SPU(gamma) i.e. pow=1:8, and infinity aSPU, based on the minimum p-values over SPU(power) each row for single SNP

Author(s)

Junghi Kim, Yun Bai and Wei Pan

References

Junghi Kim, Yun Bai and Wei Pan (2015) An Adaptive Association Test for Multiple Phenotypes with GWAS Summary Statistics, Genetic Epidemiology, 8:651-663

See Also

minP estcov

Examples

# -- n.snp: number of SNPs
# -- n.trait: number of traits
# -- n.subject: number of subjects

n.snp <- 100
n.traits <- 10
n.subjects <- 1000
traits <- matrix(rnorm(n.subjects*n.traits), n.subjects, n.traits)
v <- cov(traits)
allZ <- rmvnorm(n.snp, Sigma=v)
colnames(allZ) <- paste("trait", 1:n.traits, sep="")
rownames(allZ) <- paste("snp", 1:n.snp, sep="")


r <- estcov(allZ)
MTaSPUs(Z = allZ, v = r, B = 100, pow = c(1:4, Inf), transform = FALSE)
MTaSPUs(Z = allZ[1,], v = r, B = 100, pow = c(1:4, Inf), transform = FALSE)
minP(Zi= allZ[1,], r = r)

gene-Multitrait Sum of Powered Score (MTSPUsSet) tests and adaptive MTSPUsSet (MTaSPUsSet) test for multi trait - SNP set association with GWAS summary statistics.

Description

It gives p-values of the MTSPUsSet tests and MTaSPUsSet test with GWAS summary statistics.

Usage

MTaSPUsSet(
  Zs,
  corSNP,
  corPhe,
  pow = c(1, 2, 4, 8),
  pow2 = c(1, 2, 4, 8),
  n.perm = 5000,
  Ps = FALSE,
  prune = TRUE
)

Arguments

Value

A vector object, MTSPUsSet test P values and MTaSPUsSet P value.

Author(s)

Il-Youp Kwak and Wei Pan

References

Il-Youp Kwak, Wei Pan (2017) Gene- and pathway-based association tests for multiple traits with GWAS summary statistics, Bioinformatics, 33(1), 64-71.

Examples

data(SAMD11)
attach(SAMD11)
## example analysis using MTaSPUsSet test.
(outFZ <- MTaSPUsSet(ZsF, corSNP=corSNPF, corPhe = corPheF,
      pow=c(1,2,4,8),  pow2 = c(1,2,4,8), n.perm=10, Ps=FALSE))

Multitraits - Pathway based Sum of Powered Score tests (MTSPUsSetPath) and adaptive MTSPUsSetpath (MTaSPUsSetPath) test for multi trait - pathway association with GWAS summary statistics.

Description

It gives p-values of the MTSPUsSetPath tests and MTaSPUsSetPath test with GWAS summary statistics.

Usage

MTaSPUsSetPath(
  Zs,
  corPhe,
  corSNP,
  pow1 = c(1, 2, 4, 8),
  pow2 = c(1, 2, 4, 8),
  pow3 = c(1, 2, 4, 8),
  snp.info,
  gene.info,
  n.perm = 1000,
  Ps = FALSE,
  prune = TRUE
)

Arguments

Value

P-values for MTSPUsSetpath tests and MTaSPUsSetpPath test.

Author(s)

Il-Youp Kwak and Wei Pan

References

Il-Youp Kwak, Wei Pan (2017) Gene- and pathway-based association tests for multiple traits with GWAS summary statistics, Bioinformatics, 33(1), 64-71

See Also

MTaSPUsSet

Examples

Zs <- cbind ( c( 0.3, 0.2, 0.1,0.5,1.2),
          c(-.1, .3,-.1,.1,1.2) )

varSNP = cbind( c( 1, .1,0, 0, .11),
               c(.1,1, 0, 0, 0),
               c(0,0, 1, 0, 0),
               c(0,0, 0, 1, 0),
               c(.11,0,0,0,1) )

varPhe = cbind( c( 1, -.1),
               c(-.1,1) )

gene.info = data.frame( Gnm = c( "G1", "G2"), chr = c(1,3),
                    loc1 = c(0, 0), loc2 = c(10,10) )
snp.info = data.frame( rsid = c("rs1", "rs2", "rs3", "rs4", "rs5"),
                chr = c(1,1,3,3,3), loc = c(1,2,1,2,3) )
 out <- MTaSPUsSetPath(Zs, corPhe = varPhe, corSNP=varSNP,
            n.perm = 100, snp.info = snp.info, gene.info = gene.info)
out

gene-Multitrait Sum of Powered Score (MTSPUsSetScore) tests and adaptive MTSPUsSet (MTaSPUsSet) test for multi trait - SNP set association with GWAS summary statistics. (Score version added)

Description

It gives p-values of the MTSPUsSet tests MTScore and MTaSPUsSet_Score test with GWAS summary statistics.

Usage

MTaSPUsSetScore(
  Zs,
  corSNP,
  corPhe,
  pow = c(1, 2, 4, 8),
  pow2 = c(1, 2, 4, 8),
  n.perm = 5000,
  Ps = FALSE
)

Arguments

Value

A vector object, MTSPUsSet test, MTScore test P values and MTaSPUsSet_Score P value.

Author(s)

Il-Youp Kwak and Wei Pan

References

Il-Youp Kwak, Wei Pan (2017) Gene- and pathway-based association tests for multiple traits with GWAS summary statistics, Bioinformatics. 33(1), 64-71

Examples

data(SAMD11)
attach(SAMD11)
## example analysis using aSPUM test.
(outFZ <- MTaSPUsSetScore(ZsF, corSNP=corSNPF, corPhe = corPheF,
      pow=c(1,2,4,8),  pow2 = c(1,2,4,8), n.perm=10, Ps=FALSE))

An example data for MTaSPUsSet test

Description

Genetic Investigation of ANthropometric Traits (GIANT) consortium data contain P-values of 2.7 million SNPs with six anthropometric traits that are well established to represent body size and shape: height, weight, BMI, waist circumference (WC), hip circumference(HIP).

We mapped SNPs on gene SAMD11. This subdata contains P values, Z scores of SNPs mapped on gene SAMD11. It also contains correlation among SNPs and correlation among phenotypes for demostrated example of MTaSPUsSet test.

Usage

data(SAMD11)

References

Shungin D, Winkler TW, Croteau-Chonka DC, Ferreira T, Locke AE, Magi R, Strawbridge R, Pers TH, Fischer K, Justice AE, Workalemahu T, Wu JM, et al. (2015). New genetic loci link adipose and insulin biology to body fat distribution. Nature 518: 187-196.

Examples

data(SAMD11)

## Z-scores for man
SAMD11$ZsM

## P-values for man
SAMD11$PsM

## correlation among SNPs for man
SAMD11$corSNPM

## correlation among phenotypes for man
SAMD11$corPheM

Sum of Powered Score (SPU) tests and adaptive SPU (aSPU) test for single trait - SNP set association.

Description

It gives p-values of the SPU tests and aSPU test.

Usage

aSPU(
  Y,
  X,
  cov = NULL,
  resample = c("perm", "boot", "sim"),
  model = c("gaussian", "binomial"),
  pow = c(1:8, Inf),
  n.perm = 1000,
  threshold = 10^5
)

Arguments

Value

A list object, Ts : test statistics for the SPU tests (in the order of the specified pow) and finally for the aSPU test. pvs : p-values for the SPU and aSPU tests.

Author(s)

Il-Youp Kwak, Junghi Kim, Yiwei Zhang and Wei Pan

References

Wei Pan, Junghi Kim, Yiwei Zhang, Xiaotong Shen and Peng Wei (2014) A powerful and adaptive association test for rare variants, Genetics, 197(4), 1081-95

Junghi Kim, Jeffrey R Wozniak, Bryon A Mueller, Xiaotong Shen and Wei Pan (2014) Comparison of statistical tests for group differences in brain functional networks, NeuroImage, 1;101:681-694

See Also

aSPUw

Examples

data(exdat)

## example analysis using aSPU test on exdat data.
# increase n.perm for better p.val 
## Not run: out <- aSPU(exdat$Y, exdat$X, cov = NULL, resample = "boot",
            model = "binomial", pow = c(1:8, Inf), n.perm = 1000) 
## End(Not run)


out$Ts
# This is a vector of Test Statistics for SPU and aSPU tests.
# SPU1 to SPUInf corresponds with the option pow=c(1:8, Inf)
# They are SPU test statistics.
# The last element aSPU is minimum of them, aSPU statistic.

out$pvs
# This is a vector of p-values for SPU and aSPU tests.
# SPU1 to SPUInf corresponds with the option pow=c(1:8, Inf)
# They are p-values for corresponding SPU tests.
# The last element is p-value of aSPU test.

Adaptive Sum of powered score (SPU) tests (SPU and aSPU) based on asymptotic distribution.

Description

It gives the p-values of the SPU(1), SPU(2) and minP tests and aSPUd test based on the asymptotic distribution.

Usage

aSPUd(Y, X, cov = NULL, model = c("gaussian", "binomial"))

Arguments

Value

p-values for SPU(1), SPU(2), minP tests and aSPU test.

References

Gongjun Xu, Lifeng Lin, Peng Wei and Wei Pan (2016) An adaptive two-sample test for high-dimensional means, Biometrika (2016) 103 (3): 609-624.

See Also

aSPU

Examples

data(exdat)
out <- aSPUd(exdat$Y, exdat$X, cov = NULL, model = "binomial")
out

Pathway based Sum of Powered Score tests (SPUpath) and adaptive SPUpath (aSPUpath) test for single trait - pathway association.

Description

It gives p-values of the SPUpath tests and aSPUpath test.

Usage

aSPUpath(
  Y,
  X,
  cov = NULL,
  model = c("binomial", "gaussian"),
  snp.info,
  gene.info,
  pow = c(1:8, Inf),
  pow2 = c(1, 2, 4, 8),
  n.perm = 200,
  usePCs = F,
  varprop = 0.95
)

Arguments

Value

P-values for SPUpath tests and aSPUpath test.

Author(s)

Il-Youp Kwak and Wei Pan

References

Wei Pan, Il-Youp Kwak and Peng Wei (2015) A Powerful and Pathway-Based Adaptive Test for Genetic Association With Common or Rare Variants, The American Journal of Human Genetics, 97, 86-98

See Also

simPathAR1Snp

Examples

## Not run: dat1<-simPathAR1Snp(nGenes=20, nGenes1=5, nSNPlim=c(1, 20),
	       nSNP0=1, LOR=.2, n=100, MAFlim=c(0.05, 0.4), p0=0.05 ) 
## End(Not run)


# p-values of SPUpath and aSPUpath tests.
## Not run: p.pathaspu<- aSPUpath(dat1$Y, dat1$X, snp.info = dat1$snp.info,
         gene.info = dat1$gene.info,
         model = "binomial", pow=1:8, pow2=c(1, 2, 4, 8), n.perm=1000) 
## End(Not run)

p.pathaspu
## pow = 1:8 and pow2 = 1,2,4,8
## So, there are 8*4 = 32 SPUpath p-values.
## SPUpathi,j corresponds pow = i , pow2 = j
## The last element, aSPUpath gives aSPUpath p-value.

Pathway based Sum of Powered Score tests (SPUpath) and adaptive SPUpath (aSPUpath) test for single trait - pathway association. (vector version, fast when n is large)

Description

It gives p-values of the SPUpath tests and aSPUpath test. Faster than aSPUsPath function when n is large (N > 10^4).

Usage

aSPUpath2(
  Y,
  X,
  cov = NULL,
  model = c("binomial", "gaussian"),
  snp.info,
  gene.info,
  pow = c(1:8, Inf),
  pow2 = c(1, 2, 4, 8),
  n.perm = 200,
  usePCs = F,
  varprop = 0.95
)

Arguments

Value

P-values for SPUpath tests and aSPUpath test.

Author(s)

Il-Youp Kwak and Wei Pan

References

Wei Pan, Il-Youp Kwak and Peng Wei (2015) A Powerful and Pathway-Based Adaptive Test for Genetic Association With Common or Rare Variants, The American Journal of Human Genetics, 97, 86-98

See Also

simPathAR1Snp

Examples

## Not run: dat1<-simPathAR1Snp(nGenes=20, nGenes1=5, nSNPlim=c(1, 20),
	       nSNP0=1, LOR=.2, n=100, MAFlim=c(0.05, 0.4), p0=0.05 ) 
## End(Not run)

# p-values of SPUpath and aSPUpath tests.
## Not run: p.pathaspu<- aSPUpath2(dat1$Y, dat1$X, snp.info = dat1$snp.info,
         gene.info = dat1$gene.info,
         model = "binomial", pow=1:8, pow2=c(1, 2, 4, 8), n.perm=1000) 
## End(Not run)
#p.pathaspu
## pow = 1:8 and pow2 = 1,2,4,8
## So, there are 8*4 = 32 SPUpath p-values.
## SPUpathi,j corresponds pow = i , pow2 = j
## The last element, aSPUpath gives aSPUpath p-value.

Robust Sum of powered score (SPU) tests and aSPU test for a quantitative trait

Description

The test is based on the Huber loss function and using the parametric bootstrap for inference (i.e. bootstrapping residuals).

Usage

aSPUr(Y, X, cov = NULL, pow = c(1:8, Inf), B = 1000, C = 1.345)

Arguments

Value

p-values of the SPUr tests in the order of supplied pow values; finally, the p-value of the aSPUr test (that combines the SPUs tests with pow by taking their min P-value and adjust for multiple testing).

Author(s)

Yiwei Zhang and Wei Pan

References

Peng Wei, Ying Cao, Yiwei Zhang, Zhiyuan Xu, Il-Youp Kwak, Eric Boerwinkle, Wei Pan (2016) On Robust Association Testing for Quantitative Traits and Rare Variants, G3, 6(12) 3941-3950.

See Also

aSPU

Examples

data(exdat)

## example analysis using aSPU test on exdat data.

QT <- jitter(exdat$Y)

out <- aSPUr(Y = QT, X = exdat$X, cov = NULL, B = 100)
out
## This is a vector of p-values for SPUr and aSPUr tests.
## SPU1 to SPUInf corresponds with the option pow=c(1:8, Inf)
## They are p-values for corresponding SPUr tests.
## The last element is p-value of aSPUr test.

Sum of Powered Score (SPUs) tests and adaptive SPU (aSPUs) test for single trait - SNP set association with GWAS summary statistics.

Description

It gives p-values of the SPUs tests and aSPUs test with GWAS summary statistics.

Usage

aSPUs(Zs, corSNP, pow = c(1:8, Inf), n.perm = 1000, Ps = FALSE, prune = TRUE)

Arguments

Value

A list object, Ts : test statistics for the SPU tests (in the order of the specified pow) and finally for the aSPU test. pvs : p-values for the SPUs and aSPUs tests.

Author(s)

Il-Youp Kwak and Wei Pan

References

Il-Youp Kwak, Wei Pan (2015) Adaptive Gene- and Pathway-Trait Association Testing with GWAS Summary Statistics, Bioinformatics, 32(8), 1178-1184.

See Also

aSPUw aSPU aSPUsPath

Examples

data(kegg9)
## example analysis using aSPUM test.
g <- kegg9$gene.info[1,1]  #  SOAT1 
## Take snps mapped on gene "SOAT1" from the information of gene.info and snp.info. 
snps <- which( ( kegg9$snp.info[,2] == kegg9$gene.info[kegg9$gene.info[,1] == g, 2] ) &
                 (kegg9$snp.info[,3] > kegg9$gene.info[kegg9$gene.info[,1] == g, 3] ) &
                 (kegg9$snp.info[,3] < kegg9$gene.info[kegg9$gene.info[,1] == g, 4] )  )
## Take subsets
newP <- kegg9$nP[snps] ;
ldsub <- kegg9$ldmatrix[snps, snps];

## Get p-value for gene SOAT1. Read vignette for details.
out <- aSPUs(newP, corSNP=ldsub , pow=c(1:8, Inf), n.perm=100, Ps=TRUE)

out$Ts
# This is a vector of Test Statistics for SPUM and aSPUM tests.
# SPUs1 to SPUsInf corresponds with the option pow=c(1:8, Inf)
# They are SPUs test statistics.
# The last element aSPUs is minimum of them, aSPUs statistic.

out$pvs
# This is a vector of p-values for SPUs and aSPUs tests.
# SPUs1 to SPUsInf corresponds with the option pow=c(1:8, Inf)
# They are p-values for corresponding SPUs tests.
# The last element is p-value of aSPUs test.

Sum of Powered Score (SPUs) tests and adaptive SPU (aSPUs) test for single trait - SNP set association with GWAS summary statistics (distribution based).

Description

It gives p-values of the SPUs tests and aSPUs test with GWAS summary statistics.

Usage

aSPUsD(Zs, corrSNP, Ps = FALSE)

Arguments

Value

pvs : p-values for the SPUsD and aSPUsD tests.

Author(s)

Il-Youp Kwak and Wei Pan

References

Gongjun Xu, Lifeng Lin, Peng Wei and Wei Pan (2016) An adaptive two-sample test for high-dimensional means, Biometrika (2016) 103 (3): 609-624. Il-Youp Kwak, Wei Pan (2015) Adaptive Gene- and Pathway-Trait Association Testing with GWAS Summary Statistics, Bioinformatics, 32(8), 1178-1184

See Also

aSPUs aSPU

Examples

data(kegg9)
## example analysis using aSPUM test.
g <- kegg9$gene.info[1,1]  #  SOAT1 
## Take snps mapped on gene "SOAT1" from the information of gene.info and snp.info. 
snps <- which( ( kegg9$snp.info[,2] == kegg9$gene.info[kegg9$gene.info[,1] == g, 2] ) &
                 (kegg9$snp.info[,3] > kegg9$gene.info[kegg9$gene.info[,1] == g, 3] ) &
                 (kegg9$snp.info[,3] < kegg9$gene.info[kegg9$gene.info[,1] == g, 4] )  )
## Take subsets
newP <- kegg9$nP[snps] ;
ldsub <- kegg9$ldmatrix[snps, snps];
## Get p-value for gene SOAT1. Read vignette for details.
out <- aSPUsD(newP, corrSNP=ldsub, Ps=TRUE)

out

Pathway based Sum of Powered Score tests (SPUsPath) and adaptive SPUpath (aSPUsPath) test for single trait - pathway association with GWAS summary statistics.

Description

It gives p-values of the SPUsPath tests and aSPUsPath test with GWAS summary statistics.

Usage

aSPUsPath(
  Zs,
  corSNP,
  pow = c(1, 2, 4, 8, Inf),
  pow2 = c(1, 2, 4, 8),
  snp.info,
  gene.info,
  n.perm = 1000,
  Ps = FALSE,
  prune = TRUE
)

Arguments

Value

P-values for SPUMpath tests and aSPUMpath test.

Author(s)

Il-Youp Kwak and Wei Pan

References

Il-Youp Kwak, Wei Pan (2015) Adaptive Gene- and Pathway-Trait Association Testing with GWAS Summary Statistics, Bioinformatics, 32(8):1178-1184

See Also

aSPUs

Examples

data(kegg9)

# p-values of SPUpath and aSPUpath tests.
out.a <- aSPUsPath(kegg9$nP, corSNP = kegg9$ldmatrix, pow=c(1:8, Inf),
                  pow2 = c(1,2,4,8), 
                  snp.info=kegg9$snp.info, gene.info = kegg9$gene.info,
                  n.perm=5, Ps = TRUE)

out.a

Pathway based Sum of Powered Score tests (SPUsPath) and adaptive SPUpath (aSPUsPath) test for single trait - pathway association with GWAS summary statistics. (vector version, fast when n is large)

Description

It gives p-values of the SPUsPath tests and aSPUsPath test with GWAS summary statistics. Faster than aSPUsPath function when n is large (N > 10^4).

Usage

aSPUsPath2(
  Zs,
  corSNP,
  pow = c(1, 2, 4, 8, Inf),
  pow2 = c(1, 2, 4, 8),
  snp.info,
  gene.info,
  n.perm = 1000,
  Ps = FALSE,
  prune = TRUE
)

Arguments

Value

P-values for SPUMpath tests and aSPUMpath test.

Author(s)

Il-Youp Kwak and Wei Pan

References

Il-Youp Kwak, Wei Pan (2015) Adaptive Gene- and Pathway-Trait Association Testing with GWAS Summary Statistics, Bioinformatics, 32(8):1178-1184

See Also

aSPUs

Examples

data(kegg9)

# p-values of SPUpath and aSPUpath tests.
## Not run: out.a <- aSPUsPath2(kegg9$nP, corSNP = kegg9$ldmatrix, pow=c(1:8, Inf),
                  pow2 = c(1,2,4,8), 
                  snp.info=kegg9$snp.info, gene.info = kegg9$gene.info,
                  n.perm=1000, Ps = TRUE) 
## End(Not run)

#out.a

Inverse variance weighted Sum of Powered Score tests (SPUw) and adaptive SPUw (aSPUw) test for single trait - SNP set association.

Description

It gives the p-values of the SPUw tests and aSPUw test based on the permutations of the residuals or simulations from the null distripution.

Usage

aSPUw(
  Y,
  X,
  cov = NULL,
  resample = c("perm", "boot", "sim"),
  model = c("gaussian", "binomial"),
  pow = c(1:8, Inf),
  n.perm = 1000
)

Arguments

Value

A list object, Ts : Test Statistics for the SPUw and aSPUw test. pvs : p-values for the SPUw and aSPUw test.

Author(s)

Il-Youp Kwak, Junghi Kim and Wei Pan

References

Junghi Kim, Jeffrey R Wozniak, Bryon A Mueller, Xiaotong Shen and Wei Pan (2014) Comparison of statistical tests for group differences in brain functional networks, Neuroimage, 1;101:681-694

See Also

aSPU

Examples

data(exdat)

# increase n.perm for accurate p-value
## Not run: out <- aSPUw(exdat$Y, exdat$X, pow = c(1:8, Inf), n.perm = 1000) 


out$Ts
# This is a vector of Test Statistics for SPU and aSPU tests.
# SPU1 to SPUInf corresponds with the option pow=c(1:8, Inf)
# They are SPU test statistics.
# The last element aSPU is minimum of them, aSPU statistic.

out$pvs
# This is a vector of p-values for SPU and aSPU tests.
# SPU1 to SPUInf corresponds with the option pow=c(1:8, Inf)
# They are p-values for corresponding SPU tests.
# The last element is p-value of aSPU test.

estcov

Description

Estimate the covariance matrix of multiple traits based on their (null) summary Z-scores.

Usage

estcov(allZ, Ps = FALSE)

Arguments

Value

estimated correlation matrix.

Author(s)

Junghi Kim, Yun Bai and Wei Pan

References

Junghi Kim, Yun Bai and Wei Pan (2015) An Adaptive Association Test for Multiple Phenotypes with GWAS Summary Statistics, Genetic Epidemiology, 8:651-663

See Also

MTaSPUs minP

Examples

# -- n.snp: number of SNPs
# -- n.trait: number of traits
# -- n.subject: number of subjects

n.snp <- 100
n.traits <- 10
n.subjects <- 1000
traits <- matrix(rnorm(n.subjects*n.traits), n.subjects, n.traits)
v <- cov(traits)
allZ <- rmvnorm(n.snp, Sigma=v)
colnames(allZ) <- paste("trait", 1:n.traits, sep="")
rownames(allZ) <- paste("snp", 1:n.snp, sep="")


r <- estcov(allZ)
MTaSPUs(Z = allZ, v = r, B = 100, pow = c(1:4, Inf), transform = FALSE)
MTaSPUs(Z = allZ[1,], v = r, B = 100, pow = c(1:4, Inf), transform = FALSE)
minP(Zi= allZ[1,], r = r)

An example data set

Description

The exdat data set is list of three objects. exdat$Y is a vector of length 1000, 500 0s and 500 1s. exdat$X is a matrix of 1000 by 10. This simulated X matrix is assumed to be rare variants. All eliments are 0, 1 or 2. exdat$SNP0indx is a vector of length 10. The values are 1 or 0, 0 indicate corresponding column of X matrix have no association with Y vector.

Usage

data(exdat)

Examples

data(exdat)
exdat$X[c(1:10, 501:510),]
exdat$Y[c(1:10, 501:510)]

# 3rd and 10th column of X have no association with Y
exdat$SNP0indx

Get the p-values of the multiple SNPs based on the Wald test in a logistic regresion model.

Description

Get p-value using logistic regresion for each of the multiple SNPs

Usage

getlogitp(Y, X)

Arguments

Value

p-values for each SNPs.

See Also

GATES2 GatesSimes Hyst

Examples

#simula <- simPathAR1Snp(nGenes=20, nGenes1=1, nSNPlim=c(1, 20), nSNP0=1:3,
#                           LOR=.2, rholim=c(0,0),
#                           n=10, MAFlim=c(0.05, 0.4), p0=0.05)
#logitp <- getlogitp(simula$Y, simula$X)

A subset of coronary artery disease (CAD) data

Description

Data on coronary artery disease, myocardial infarction have been contributed by CARDIoGRAMplusC4D investigators and have been downloaded from www.CARDIOGRAMPLUSC4D.ORG

The data set contains P value data for coronary artery disease (CAD). We mapped these SNPs to the 9th KEGG pathway.

gene.info is a contains 16 gene information of 9th Kegg pathway. The 1st column is the name of the gene, 2nd column is the chromosome number, 3rd column is where the gene starts and 4th column is where the gene ends.

snp.info contains 330 snp informations mapped on this Kegg pathway. The 1st column is the rsID of the SNP, 2nd column is the chromosome number and the 3rd column is the location of the SNP.

Ps is a vector containing p-value information for mapped SNPs.

The 'PPs' is a list object contains the SNP information for each genes. For example, PPs$CEL contains snp.info matrix mapped on gene 'CEL'. The 1st column is the rsID of the SNP, 2nd column is the chromosome number, the 3rd column is the location of the SNP and the 4th column is p-value of the SNP.

nP is a subvector of Ps. These are SNPs on reference population(Hapmap CEU phase 2 data, downloadable from plink, SNPs with MAF less than 5 percent omitted)

ldmatrix is a correaltion matrix of SNPs from the reference population.

Usage

data(kegg9)

References

Schunkert H, Konig IR, Kathiresan S, Reilly MP, Assimes TL, Holm H et al. (2011) Large-scale association analysis identifies 13 new susceptibility loci for coronary artery disease. Nat Genet. 43: 333-338

Examples

data(kegg9)

## gene informations
kegg9$gene.info

## SNPs mapped on 3rd and 4th gene in 9th Kegg pathway
kegg9$PPs[3:4]

## snp information
kegg9$snp.info

## The 1st 10 P-values of SNPs mapped on 9th Kegg pathway. 
kegg9$nP[1:10]

## correlation matrix among those SNPs
kegg9$ldmatrix[1:10,1:10]

minP test.

Description

Return exact minP test p-value for multiple traits - single SNP association.

Usage

minP(Zi, r)

Arguments

Value

return exact minP test

Author(s)

Junghi Kim, Yun Bai and Wei Pan

References

Junghi Kim, Yun Bai and Wei Pan (2015) An Adaptive Association Test for Multiple Phenotypes with GWAS Summary Statistics, Genetic Epidemiology, 8:651-663

See Also

estcov MTaSPUs

Examples

# -- n.snp: number of SNPs
# -- n.trait: number of traits
# -- n.subject: number of subjects

n.snp <- 100
n.traits <- 10
n.subjects <- 1000
traits <- matrix(rnorm(n.subjects*n.traits), n.subjects, n.traits)
v <- cov(traits)
allZ <- rmvnorm(n.snp, Sigma=v)
colnames(allZ) <- paste("trait", 1:n.traits, sep="")
rownames(allZ) <- paste("snp", 1:n.snp, sep="")


r <- estcov(allZ)
MTaSPUs(Z = allZ, v = r, B = 100, pow = c(1:4, Inf), transform = FALSE)
MTaSPUs(Z = allZ[1,], v = r, B = 100, pow = c(1:4, Inf), transform = FALSE)
minP(Zi= allZ[1,], r = r)

Image plot of P-value matrix.

Description

It gives a P-value image for a given P-value matrix.

Usage

plotPmat(
  Ps,
  zlim = NULL,
  main = NULL,
  yt = NULL,
  xlab = "SNPs",
  thresh = -log(5e-08, 10),
  trait.names = NULL
)

Arguments

Value

Image of P-values matrix.

Author(s)

Il-Youp Kwak and Wei Pan

References

Il-Youp Kwak, Wei Pan (2017) Gene- and pathway-based association tests for multiple traits with GWAS summary statistics, Bioinformatics. 33(1), 64-71

See Also

aSPUs

Examples

## Say we have 3 traits and their p-values at 5 SNPs. 
Ps <- rbind( c(0.001, 0.4, 0.5, 0.00000001, .1),
           c(0.03, 0.3, 0.3, 0.00001, .2),
           c(0.01, 0.2, 0.4, 0.001, .0001) )

## We can visualize it using plotPmat function.
plotPmat(Ps)

Prune SNPs (related to aSPUs, aSPUsPath, MTaSPUsSet)

Description

When correlation matirx have highly correlated SNPs, the performance of aSPUs, aSPUsPath and MTaSPUsSet are not very good. Do pruning using this function.

Usage

pruneSNP(corSNP, rup = 0.95, rdown = -0.95)

Arguments

Value

a list object pruned.corSNP and to.erase. pruned.corSNP is pruned correlation matrix. to.erase is SNP index to erase to get purned object. (i.e. corSNP[-to.erase, -to.erase] = pruned.corSNP )

Author(s)

Il Youp Kwak

See Also

aSPUs aSPUsPath MTaSPUsSet

Examples

data(kegg9)
## example analysis using aSPUM test.
g <- kegg9$gene.info[1,1]  #  SOAT1 
## Take snps mapped on gene "SOAT1" from the information of gene.info and snp.info. 
snps <- which( ( kegg9$snp.info[,2] == kegg9$gene.info[kegg9$gene.info[,1] == g, 2] ) &
                 (kegg9$snp.info[,3] > kegg9$gene.info[kegg9$gene.info[,1] == g, 3] ) &
                 (kegg9$snp.info[,3] < kegg9$gene.info[kegg9$gene.info[,1] == g, 4] )  )
## Take subsets
newP <- kegg9$nP[snps] ;
ldsub <- kegg9$ldmatrix[snps, snps];

prSNP <- pruneSNP(ldsub) 
newP <- newP[-prSNP$to.erase]
ldsub <- ldsub[-prSNP$to.erase, -prSNP$to.erase ]

## Get p-value for gene SOAT1. Read vignette for details.
out <- aSPUs(newP, corSNP=ldsub , pow=c(1:8, Inf), n.perm=100, Ps=TRUE)

out$Ts
# This is a vector of Test Statistics for SPUM and aSPUM tests.
# SPUs1 to SPUsInf corresponds with the option pow=c(1:8, Inf)
# They are SPUs test statistics.
# The last element aSPUs is minimum of them, aSPUs statistic.

out$pvs
# This is a vector of p-values for SPUs and aSPUs tests.
# SPUs1 to SPUsInf corresponds with the option pow=c(1:8, Inf)
# They are p-values for corresponding SPUs tests.
# The last element is p-value of aSPUs test.

Simulating a pathway with multiple SNPs.

Description

It gives a simulated SNPs consisting of multiple genes in a pathway. Each SNPs from a latent multivariate Gaussian variable with an AR1 correlation structure.

Usage

simPathAR1Snp(
  nGenes = 10,
  nGenes1 = 5,
  nSNPs = NULL,
  ncSNPs = NULL,
  nSNPlim = c(1, 20),
  nSNP0 = 1:3,
  LOR = 0.3,
  n = 100,
  MAFlim = c(0.05, 0.4),
  rholim = c(0, 0),
  p0 = 0.05,
  noncausal = FALSE
)

Arguments

Value

a list of the binary outcome Y (=0 or 1) and SNPs (=0, 1 or 2); Y is a vector of length 2n; X is a matrix of 2n by nSNP.

See Also

aSPUpath

Examples

# Simulation set up A a) in the paper (Pan et al 2015)
## Not run:  simula <- simPathAR1Snp(nGenes=20, nGenes1=1, nSNPlim=c(1, 20), nSNP0=1:3,
                           LOR=.2, rholim=c(0,0),
                           n=100, MAFlim=c(0.05, 0.4), p0=0.05) 
## End(Not run)


# Simulation set up A b) in the paper
#simulb <- simPathAR1Snp(nGenes=20, nGenes1=1, nSNPlim=c(1, 100), nSNP0=1:3,
#                           LOR=.2, rholim=c(0,0),
#                           n=100, MAFlim=c(0.05, 0.4), p0=0.05)

An example data for MTaSPUsSet test

Description

Genetic Investigation of ANthropometric Traits (GIANT) consortium data contain P-values of 2.7 million SNPs with six anthropometric traits that are well established to represent body size and shape: height, weight, BMI, waist circumference (WC), hip circumference(HIP).

We mapped SNPs on gene LCORL, RASA2, STK33 and RPGRIP1L. This subdata contains P values, correlation matrices for demostrated example of MTaSPUsSet test.

Usage

data(someGs)

References

Shungin D, Winkler TW, Croteau-Chonka DC, Ferreira T, Locke AE, Magi R, Strawbridge R, Pers TH, Fischer K, Justice AE, Workalemahu T, Wu JM, et al. (2015). New genetic loci link adipose and insulin biology to body fat distribution. Nature 518: 187-196.

Examples

data(someGs)

## P-values for LCORL
someGs$LCORL[[1]]

## correlation matrix for LCORL
someGs$LCORL[[2]]

## P-values for RASA2
someGs$RASA2[[1]]

## correlation matrix for LCORL
someGs$RASA2[[2]]

# See vignettes for more details.