run.SHLR.scan: SHLR scan

In SHLR: Shared Haplotype Length Regression

Description

This is a wrapper function that performs SHLR or SHLR-fam statistic to a genomic region one marker at a time. The scan results will be outputted to a file.

Usage

run.SHLR.scan(haps, phen, marker.map, outfile, outcome, 
out.dist = "Gaussian", cov = "", missing.id.code = "0", 
method, nSplits = 1, pca.thres = NULL, hap.freq.thres = 0.005, 
window.size = 100, threads = NULL, corstr = "kinship", 
std.err = "naive", gender = "", fid = "", iid = "", pid = "", 
mid = "")

Arguments

haps	Haplotypes array stored as a string of character per row. Every two rows contains the haplotypes for each individual. (Required)
phen	Phenotype table. The phenotype table contains information about trait of interest and other covariates. (Required)
marker.map	The marker map contains information about marker name (1st column) and marker position (2nd column). (Required)
outfile	The desired name of the output file. (Required)
outcome	The column name for the phenotype of interest in the phenotype table. (Required)
out.dist	The assumed outcome distribution of the model. e.g. : gaussian, binomial, etc. (Required)
cov	The column names for the covariates in the phenotype table.
missing.id.code	The code for missing id in the pedigree structure. Usually, it is coded as '0'. (Required only if method is 'SHLR-fam' and corstr is 'kinship')
method	The choice for the SHLR method (i.e. : SHLR or SHLR-fam). (Required)
nSplits	The number of splits. This option will determine how often the program prints out results to the output file.
pca.thres	The proportion of variation retained in the design matrix of the model based on PCA.
hap.freq.thres	The minimum haplotype frequency proportion threshold. Haplotypes that meet this threshold will be considered in the model.
window.size	The number of markers that form the haplotype.
threads	The number of CPUs for parallel computation.
corstr	The type of correlation structure. The options are: 'kinship', 'exchangeable', or 'independence'. (SHLR-fam parameter)
std.err	The method for estimating variance for SHLR-fam. The options are: 'naive' or 'sandwich'. (SHLR-fam parameter)
gender	The column name that indicates gender in the phenotype table. (Required if method is SHLR-fam and cor.str is 'kinship')
fid	The column name that indicates family id in the phenotype table. (Required if method is SHLR-fam and cor.str is 'kinship')
iid	The column name that indicates individual id in the phenotype table. (Required if method is SHLR-fam and cor.str is 'kinship')
pid	The column name that indicates father id in the phenotype table. (Required if method is SHLR-fam and cor.str is 'kinship')
mid	The column name that indicates mother id in the phenotype table. (Required if method is SHLR-fam and cor.str is 'kinship')

Examples

library(SHLR)

####reading in pedigree data
fam <- read.table(system.file("extdata","plink.fam",package="SHLR"), 
    header=FALSE, stringsAsFactors=FALSE)
colnames(fam) <- c("fam_id","id","dad_id","mom_id","sex","qtrait")

####add simulated covariates to create phenotype table
N <- dim(fam)[1]
cov1 <- rnorm(N, 0, 2)
cov2 <- rbinom(N, 1, 0.7)
phen <- data.frame(fam, cov1=cov1, cov2=cov2)

##Read phased haplotype
shapeit <- read.shapeit.haps(system.file("extdata","phased.haps",package="SHLR"))
haps <- shapeit$haps
marker.map <- shapeit$marker.map

##Initialize parameters for SHLR.scan
outfile <- "out.txt"
outcome <- "qtrait"
cov <- c("sex","cov1","cov2")
famid <- "fam_id"
iid <- "id"
pid <- "dad_id"
mid <- "mom_id"
gender <- "sex"
missid <- "0"
corstr <- "kinship"
out_dist <- "gaussian"
std.err <- "naive"
method <- "SHLR-fam"
window.size <- 35
pca.thres <- 0.8
nSplits <- 1

run.SHLR.scan(haps, phen, marker.map, outfile,
        outcome, out.dist=out_dist, cov=cov, missing.id.code=missid,
        method=method, nSplits=nSplits, pca.thres=pca.thres, hap.freq.thres=0.001,
        window.size=window.size, threads=2, corstr=corstr,
        std.err=std.err, gender=gender, fid=famid, iid=iid, pid=pid, mid=mid)

SHLR documentation built on May 30, 2017, 1:23 a.m.