23andme-impute

Scripts and advice to run IMPUTE2 on 23 and me raw data

Usage:

impute_genome.pl -i input_file [-o output_prefix -s ] [impute_options]

Imputes whole genome SNPs from the raw data of ~450 000 SNPs typed by 23andme, and also phases the typed sites. This is a computationally heavy task, and so running output commands in parallel is recommended.

input_options

-i, --input The 'raw data' file from 23 and me, which has four tab separated columns: rsid, chromosome, position, genotype

-o, --output The prefix for the output .gen files (default is the input name)

-s, --sex The sex of the subject, either male (m) or female (f). If omitted a guess will be made based on presence of Y chromosome sites

impute_options:

-r, --run [threads] Directly execute the imputation, but note this requires a lot of memory and CPU time. Optionally supply an integer number of jobs to simultaneously execute. This should be <= number of cores, but this is not checked

-p, --print Print the imputation commands to STDOUT rather than executing them - default behaviour

-w, --write Write to commands to shell scripts, for execution later by a job scheduling system

other

-h, --help Displays this help message

For example:

./impute_genome.pl -i 23andme_rawdata.txt -o imputed -s m -r 4

Would impute the genome from 23andme_rawdata.txt, which is a male subject (homozygous X chromosome, has Y chromosome data) and then runs the analysis in parallel over four separate threads