snpScreen {GGtools}R Documentation

compute model fits over a sequence of SNPs

Description

compute model fits over a sequence of SNPs

Usage

snpScreen(racExSet, snpMeta, gene, formTemplate, fitter, gran, ...)
extract_p(ssr)
plot_mlp(ssr, snpMeta, gchr=NULL, geneLocDF=NULL, ps = NULL, pch = 20, cex = 0.5, local=FALSE, 
  plotf=smoothScatter, organism="human")

Arguments

racExSet instance of racExSet-class
snpMeta instance of snpMeta-class
gene instance of genesym-class class identifying the expression phenotype to be regarded as dependent variable
gchr string stating chromosome on which gene is resident; looked up in geneLocDF if not supplied
geneLocDF data frame that has one row for each gene symbol, in column `gene', and the chromosome on which it lives, in column `chr'
formTemplate a formula having form ~. or ~factor(.), literally, to specify additive or nonadditive models for effects of rare allele copy number
fitter R fitting function that can work with formulas and data frames, for example lm, or, if not a fitting function, a special function in the package, either fastHET, or fastAGM, see details below
gran a numeric convenience parameter; if not 1, SNPs will be deterministically selected with frequency 1/gran along the linear sequence in the racExSet racAssay structure
... ... – not in use
ssr snpScreenResult instance
ps a string with a probe set identifier
pch for passage to scatterSmooth
cex for passage to scatterSmooth
local by default (local==FALSE) the plot is made over the entire chromosome; if local==TRUE, the plot is made over the segment of the chromosome within which the snps screened lie; this only makes a difference if the racExSet used in the screen has been SNP-filtered relative to the HapMap SNP set.
plotf function to use for rendering – with really high-density genotyping, the default works well; in sparse cases, use plot.
organism string used for plot annotation

Details

for snpScreen: If options()$verbose == TRUE then every 100th index in the vector of snps is printed to stdout to show rate of progress.

snpScreenResult is a container for relevant information about a screen, including a list of fit objects.

For result processing, many SNPs have no variation in observed samples and statistical tests of association are indeterminate. NAs will be returned for tests on such SNPs.

fastAGM is a C routine for simple least square fitting of an additive genetic model. fastHET will compare heterozygous to homozygous.

plot_mlp returns a list of x and y values for the plotted points.

Value

creates a list of model fit results; try is used to allow failure of fit (e.g., lm may fail if a singular model matrix is computed

Author(s)

Vince Carey <stvjc@channing.harvard.edu>

Examples

example(make_racExSet)
dem
dem = exclMono(dem)
snpNames(dem)[1:4]
featureNames(dem)[1:4]
data(chr20GGdem)
data(chr20meta)
data(geneLocs_hsa)
scr1 = snpScreen(dem, chr20meta, genesym("DDR1"), ~., lm,  gran=100 )
scr1[[1]]
scr2 = snpScreen(dem, chr20meta, genesym("DDR1"), ~factor(.), lm,  gran=200 )
scr2[[1]]
plot_mlp(scr1, chr20meta, geneLocDF=geneLocs_hsa)
chr20GGdem = exclMono(chr20GGdem)
ut = unix.time(scr2 <- snpScreen(chr20GGdem, chr20meta, genesym("CPNE1"), ~factor(.), fastAGM, 50))
ut
scr2
plot_mlp(scr2, chr20meta, geneLocDF=geneLocs_hsa)
#
# here we work on a WebQTL computation
#
# get the expr+genotype data
data(gse2031GG)
# get a provisional snp metadata structure
data(INB34snpMeta)
# run a screen for Erdr1
ss = snpScreen(gse2031GG, INB34snpMeta, genesym("Erdr1"), ~., fastAGM, 1)
plot_mlp(ss, INB34snpMeta, gchr="all", plotf=plot, organism="mouse" )

[Package GGtools version 1.4.0 Index]