seqPCA: Run PCA-seq on a GDS file
In jellily/PCAseq: Population Structure Inference with Whole Genome Sequence Data

Description Usage Arguments Details Value Note

This function calculates the Genetic Relatedness Matrix (GRM) on a GDS file using the PCA-seq method for sequence data.

seqPCA(gdsobj, weights = c(1, 1), sample.id = NULL, snp.id = NULL,
  autosome.only = TRUE, remove.monosnp = TRUE, maf = NaN,
  missing.rate = NaN, eigen.cnt = 32, need.genmat = FALSE,
  verbose = TRUE)

`gdsobj`	an object of the class SNPGDSFileClass, a SNP GDS file.
`weights`	a vector of two numbers, indicating the paraters to use for the beta function weights; see Details.
`sample.id`	a vector of sample ids specifying the samples to use for analysis; if NULL, all samples are used.
`snp.id`	a vector of SNP ids specifying the SNPs to use for analysis; if NULL, all SNPs are used.
`autosome.only`	if TRUE, use autosomal SNPs only; if it is a numeric or character vector, keep SNPs according to the specified chromosomes.
`remove.monosnp`	if TURE, remove monomorphic SNPs.
`maf`	if one number is specified, use SNPs with MAF greater than or equal to this value; if a numeric vector of length two is specified, only SNPs with MAFs in (min, max) are taken.
`missing.rate`	to use the SNPs with missing rates less than or equal to missing.rate; if NaN, no misisng threshold.
`eigen.cnt`	the number of eigen vectors and values to return; if zero, return all eigenvalues and vectors.
`need.genmat`	if TRUE, return the genetic relatedness matrix.
`verbose`	Not supported.

If method is "eigen", the GRM is calculated using the EIGENSTRAT method as given in Patterson et al 2006. If method is "pcaseq", the GRM is calculated using the PCA-seq method.

Return a snpPCAClass object, a list with the follow slots:

weights: the parameters used to define the weights used to calculate the GRM
maf: the MAF cutoffs used
sample.id: the sample ids used in the analysis
snp.id: the SNP ids used in the analysis
eigenval: eigenvalues
eigenvect: a matrix of eigenvectors of dimensions # of samples by eigen.cnt
varprop: the proportion of the variance explained by each principal component
TraceXTX: the trace of the genetic relateness matrix
Bayesian: indicates Bayes normalization; set to FALSE, as this is not currently supported
genmat: the genetic relateness matrix

If you need to run the EIGENSTRAT method on a very large data set and do not need to subset by both a minimum and maximum MAF, the snpgdsPCA function will be faster.

jellily/PCAseq documentation built on May 19, 2019, 4:02 a.m.