snpgdsMergeGRM {SNPRelate} | R Documentation |
Combine multiple genetic relationship matrices with weighted averaging.
snpgdsMergeGRM(filelist, out.fn=NULL, out.prec=c("double", "single"), out.compress="LZMA_RA", weight=NULL, verbose=TRUE)
filelist |
a character vector, list of GDS file names |
out.fn |
NULL, return a GRM object; or characters, the output GDS file name |
out.prec |
double or single precision for storage |
out.compress |
the compression method for storing the GRM matrix in the GDS file |
weight |
NULL, weights proportional to the numbers of SNPs; a numeric vector, or a logical vector (FALSE for excluding some GRMs with a negative weight, weights proportional to the numbers of SNPs) |
verbose |
if |
The final GRM is the weighted averaged matrix combining multiple GRMs. The merged GRM may not be identical to the GRM calculated using full SNPs, due to missing genotypes or the internal weighting strategy of the specified GRM calculation.
None or a GRM object if out.fn=NULL
.
Xiuwen Zheng
# open an example dataset (HapMap) genofile <- snpgdsOpen(snpgdsExampleFileName()) snpid <- read.gdsn(index.gdsn(genofile, "snp.id")) snpid <- snpid[snpgdsSNPRateFreq(genofile)$MissingRate == 0] # there is no missing genotype grm <- snpgdsGRM(genofile, snp.id=snpid, method="GCTA") # save two GRMs set1 <- grm$snp.id[1:(length(grm$snp.id)/2)] set2 <- setdiff(grm$snp.id, set1) snpgdsGRM(genofile, method="GCTA", snp.id=set1, out.fn="tmp1.gds") snpgdsGRM(genofile, method="GCTA", snp.id=set2, out.fn="tmp2.gds") # merge GRMs and export to a new GDS file snpgdsMergeGRM(c("tmp1.gds", "tmp2.gds"), "tmp.gds") # return the GRM grm2 <- snpgdsMergeGRM(c("tmp1.gds", "tmp2.gds")) # check f <- openfn.gds("tmp.gds") m <- read.gdsn(index.gdsn(f, "grm")) closefn.gds(f) summary(c(m - grm$grm)) # ~zero summary(c(m - grm2$grm)) # zero # close the file snpgdsClose(genofile) # delete the temporary file unlink(c("tmp1.gds", "tmp2.gds", "tmp.gds"), force=TRUE)