RDocumentation: read.vcf

Note that this notebook was automatically generated from an RDocumentation page. It depends on the package and the example code whether this code will run without errors. You may need to edit the code to make things work.

if(!require('pegas')) {
    install.packages('pegas')
    library('pegas')
}

if (FALSE) {
## Chr Y from the 1000 Genomes:
a <- "https://ftp-trace.ncbi.nih.gov/1000genomes/ftp/release/20130502"
b <- "ALL.chrY.phase3_integrated_v1b.20130502.genotypes.vcf.gz"
## WARNING: the name of the file above may change
url <- paste(a, b, sep = "/")
## Solution 1: download first
download.file(url, "chrY.vcf.gz")
## no need to uncompress:
(info <- VCFloci("chrY.vcf.gz"))
str(info) # show the modes of the columns
## Solution 2: read remotely (since pegas 0.11)
info2 <- VCFloci(url)
identical(info, info2)
rm(info2)

SNP <- is.snp(info)
table(SNP) # how many loci are SNPs?
## compare with:
table(getINFO(info, "VT"))

op <- par(mfcol = c(4, 1), xpd = TRUE)
lim <- c(2.65e6, 2.95e6)
## distribution of SNP and non-SNP mutations along the Y chr:
plot(info$POS, !SNP, "h", col = "red", main = "non-SNP mutations",
     xlab = "Position", ylab = "", yaxt = "n")
rect(lim[1], -0.1, lim[2], 1.1, lwd = 2, lty = 2)
plot(info$POS, SNP, "h", col = "blue", main = "SNP mutations",
     xlab = "Position", ylab = "", yaxt = "n")
rect(lim[1], -0.1, lim[2], 1.1, lwd = 2, lty = 2)
par(xpd = FALSE)
## same focusing on a smaller portion of the chromosome:
plot(info$POS, !SNP, "h", col = "red", xlim = lim, xlab = "Position",
     ylab = "", yaxt = "n")
plot(info$POS, SNP, "h", col = "blue", xlim = lim, xlab = "Position",
     ylab = "", yaxt = "n")
par(op)

## read both types of mutations separately:
X.SNP <- read.vcf("chrY.vcf.gz", which.loci = which(SNP))
X.other <- read.vcf("chrY.vcf.gz", which.loci = which(!SNP))

identical(rownames(X.SNP), VCFlabels("chrY.vcf.gz")) # TRUE
cat(VCFheader("chrY.vcf.gz"))

## get haplotypes for the first 10 loci:
h <- haplotype(X.SNP, 1:10)
## plot their frequencies:
op <- par(mar = c(3, 10, 1, 1))
plot(h, horiz=TRUE, las = 1)
par(op)
}