# Multi-Locus Linkage Disequilibrium `LDscan` computes a matrix of pairwise linkage disequilibrium (LD) coefficients ($|r|$) from a set of loci (which must be bi-allelic; if not, the results are not guaranteed to be meaningful). The genotypes must be phased. `LDmap` plots a matrix of LD coefficients, optionally with the positions of the loci. ```r LDscan(x, ...) ## S3 method for class 'DNAbin' LDscan(x, quiet = FALSE, what = c("r", "Dprime"), ...) ## S3 method for class 'loci' LDscan(x, depth = NULL, quiet = FALSE, what = c("r", "Dprime"), ...) LDmap(d, POS = NULL, breaks = NULL, col = NULL, border = NA, angle = 0, asp = 1, cex = 1, scale.legend = 0.8, ...) ``` ## Arguments - `x`: an object of class `"loci"` with phased genotypes. - `depth`: a vector of integers giving the the depth(s) (or lags) at which the $r$'s are calculated. By default, all possible depths are considered. - `quiet`: a logical: should the progress of the operation be printed? - `what`: the quantity to be computed. Two choices are possible: `"r"` (the default) for the absolute value of the correlation between alleles and `"Dprime"` for the (scaled) coefficients. - `d`: a correlation matrix (can be an object of class `"dist"`). - `POS`: an optional vector of locus positions (e.g., from a VCF file; see examples). - `breaks`: a vector of break intervals to count the values in `d`; by default, ten equally-sized intervals are used. - `col`: an optional vector of colours; a scale from lightyellow to red is used by default. - `border`: the border of the rectangles: the default is to have no border (this is not the same than default in `rect`; see examples). - `angle`: value (in degrees) to rotate the graphic. - `asp`: the aspect ratio of the graphic; one by default so the elements are squares (not rectangles). - `cex`: the scaling of the labels and text. - `scale.legend`: the scaling of the legend rectangles. - ``...``: further arguments passed to methods (`LDscan`) or to `plot.default` (`LDmap`). ## Details The LD coefficient $r$ is well defined when the two loci have only two alleles. In other cases, LD is well defined (see `LD`) but the definition of $r$ is not clear. All levels of ploidy are accepted, but all loci should have the same ploidy level. If `depth` is used, the $r$'s are calculated only for the pairs of loci that are distant by these values in `x`, but necessarily on the chromosome. The returned list has names set with the values of `depth`. The value returned is actually $|r|$ (not $r^2$). ## Returns `LDscan` returns an object of class `"dist"` by default, or a list if `depth` is used. ## Author(s) Emmanuel Paradis ## See Also `LD`, `read.vcf` ## Examples ```r data(woodmouse) d <- LDscan(woodmouse) LDmap(d, seg.sites(woodmouse), seq(0, 1, .1)) ## Not run: ## Download the VCF file from Dryad: ## https://doi.org/10.5061/dryad.446sv.2 ## the VCF file should have this name: fl <- "global.pop.GATK.SNP.hard.filters.V3.phased_all.pop.maf.05.recode.vcf.gz" info.fly <- VCFloci(fl) ## LD map from the first 100 loci: x <- read.vcf(fl, to = 100) # read only 100 loci res <- LDscan(x) bks <- seq(0, 1, 0.2) LDmap(res, info.fly$POS[1:100], bks, scale.legend = 3) ## check the chromosomes: table(info.fly$CHROM) ## LD map from 100 loci randomly distributed on the chromosome: s <- ceiling(seq(1, 224253, length.out = 100)) xs <- read.vcf(fl, which.loci = s) res2 <- LDscan(xs) LDmap(res2, info.fly$POS[s], bks, scale.legend = 3) ## something simpler with 10 loci: x10 <- x[, 1:10] ## the VCF file has no locus IDs, so we give some here: names(x10) <- paste0("Loc", 1:10) res10 <- LDscan(x10, quiet = TRUE) LDmap(res10, angle = 45, border = NULL) ## End(Not run) ```

LDscan function