RZooRoH (version 0.3.2.1)

zoodata: Read the genotype data file

Description

Read a data file and convert it to the RZooRoH format in a 'zooin' object required for further analysis.

Usage

zoodata(
  genofile,
  min_maf = 0,
  zformat = "gt",
  supcol = 5,
  chrcol = 1,
  poscol = 3,
  allelefreq = NULL,
  samplefile = NA
)

Value

The function return a zooin object called containing the following elements: zooin@genos a matrix with the genotypes or genotype probabilities, zooin@bp an array with marker positions, zooin@chrbound a matrix with the first and last marker number for each chromosome, zooin@nind the number of individuals, zooin@nsnps the number of markers conserved after filtering for minor allele frequency, zooin@freqs an array with the marker allele frequencies, zooin@nchr the number of chromosomes, zooin@zformat the format of the data ("gt","gp","gl","ad") and zooin@sample_ids (the names of the samples).

Arguments

genofile

The name of the input data file. Note that the model is designed for autosomes. Other chromosomes and additional filtering (e.g. call rate, missing, HWE, etc.) should be performed prior to run RZooRoH with tools such as PLINK or bcftools for instance. The model works on an ordered map and ignores SNPs with a null position.

min_maf

The minimum allele frequency to keep variants in the analysis (optional / set to 0.00 by default to keep all markers). Values such as 0.01 allows to exclude monomorphic markers that are not informative and to reduce the size of the data and computational costs. There is no marker exclusion on call rate. However, we expect that data filtering is done prior to RZooRoH with tools such as PLINK or vcftools.

zformat

The code corresponding to the format of the data file ("gt" for genotypes, "gp" for genotype probabilities, "gl" for genotype likelihoods in Phred scores, "ad" for allelic depths). For all these formats, markers are ordered per rows and individuals per columns. Variants should be ordered by chromosome and position. By default, the first five columns are chromosome identification (e.g, "1", "chr1"), the name of the marker, the position of the marker in base pairs or better in cM multiplied by 1,000,000 when genetic distances are known, the first marker allele and the second marker allele. Information per individual varies according to the format. With the "gt" format we have one column per individual with 0, 1 and 2 indicating the number of copies of the first allele (and 9 for missing). With the "gp" format we have three column per individual with the probabilities of genotype 11 (homozygous for the first allele), genotype 12 and genotype 22 (this corresponds to the oxford GEN format). Similarly, with the "gl" format, we have three column per individual with the likelihoods for genotypes 11, 12 and 22 in Phred scores. Finally, with the "ad" format, we expect two columns per individual: the number of reads for allele 1 and the number of reads for allele 2. For these three last formats, missing values must be indicated by setting all elements to 0. If one of the columns is non-null for one individual, the genotype will be considered non-missing. Note that the marker alleles specified in columns 4 and 5 are not used.

Conversion of a PLINK ped file or a VCF file to RZooRoH format can easily be performed using PLINK (version 1.9) or using bcftools.

For ped files, recode them to oxford gen format with plink --file myinput --recode oxford --autosome --out myoutput. The autosome option keeps only SNPs on autosomes as required by RZooRoH.

For vcf files, bcftools can be used to recode a vcf to the oxford gen format with the convert option: bcftools convert -t ^chrX,chrY,chrM -g outfile --chrom --tag GT myfile.vcf. The --chrom option is important to obtain chromosome number in the first column. The tag option allows to select which field from the vcf file (GT, PL, GL or GP) is used to generate the genotype probabilities exported in the oxford gen format. The -t option allows to exclude chromosomes (this is an example and chromosome names must be adapted if necessary). The needed output data is then outfile.gen.

If some genotype probabilities are missing, with a value of "-nan", you must replace them with "0" (triple 0 is considered as missing). This can be done with this command:

sed -e 's/-nan/0/g' file.gen > newfile.gen

supcol

An optional argument that indicates the number of additional columns before the individuals genotypes (five columns are expected by default as described in the zformat argument description). Note that the function requires at least two information columns: the chromosome number and the marker position.

chrcol

An optional argument that indicates the column number where the chromosome information is indicated (first column by default).

poscol

An optional argument that indicates the column number where the marker position is indicated (third column by default).

allelefreq

A vector with allele frequencies for the first marker allele (optional). By default, the allele frequencies are estimated from the data. The option allows to skip this computation or to provide external allele frequencies estimated by another method or on another data set.

samplefile

A file with names of the samples (optional). It must match with the number of genotypes. If none is provided, the position in the genofile is used as ID.

Examples

# Get the name and location of example files

myfile1 <- system.file("exdata","genoex.txt",package="RZooRoH")
myfile2 <- system.file("exdata","genosim.txt",package="RZooRoH")

# Load your data with default format into a zooin object named "data1":

data1 <- zoodata(myfile1)

# Load the first data file with default format and filtering out markers with MAF < 0.02
# into a zooin object called "data1frq002":

data1frq002 <- zoodata(myfile1, min_maf = 0.02)

# Load the first data file with default format, with external allele frequencies
# (here a random set we create) and filtering out markers with MAF < 0.01:

myrandomfreq <- runif(14831)
data1c <- zoodata(myfile1, allelefreq = myrandomfreq, min_maf = 0.01)

# Load the second data file and indicate your own format (chromosome number in column 1,
# map position in column 2, 4 columns before genotypes) and filtering out markers with
# MAF < 0.01. The created zooin object is called "Sim5":

Sim5 <- zoodata(myfile2, chrcol = 1, poscol =2, supcol = 4, min_maf = 0.01)