Frequently Asked Questions in GIGI


1.How do I choose framework markers?

Because the framework markers are assumed to be sparse, we want to choose framework markers that are informative about which chromosomes are being transmitted at the framework loci. First, framework markers typed on a large number of subjects tend to be most informative. Second, framework markers that are multi-allelic tend to be more informative than di-allelic markers if they are available. Third, if framework markers are SNPs, markers with high minor allele frequencies in the sample tend to be more informative. Fourth, framework markers should be moderately but not too dense (eg. not denser than 1 marker per 0.3 cM because of concern about MCMC mixing and violation of the assumption of Linkage Equilibrium) If the framework markers are multiallelic, this spacing should be greater, e.g., not denser than 1 marker per 2 cM for good MCMC mixing.

2.What do I have to be careful about the genetic map that I am using?

The map positions are in map distances based on the Haldane map function, instead of Kosambi map function or sequence positions. If this map is based on the Kosambi map function, the user will need to convert the map positions to Haldane map function using an appropriate conversion method. Also, since recombination fractions are relative to each other, we strongly encourage the user to generate both the framework map positions and the dense marker positions at the same time.

3.Can GIGI process multiple pedigrees?

No. In version 1.01, GIGI can only process 1 pedigree at a time. For this reason, gl_auto should also be run individually for each pedigree.

4.Can I use GIGI to impute genotypes on unrelated individuals?

No.

5.What should I know when I use GIGI to impute genotypes on trios?

Since there is no way to infer recombination on trios using unphased genotypes, gl_auto produces an empty inheritance vectors file. From the pedigree-meiosis file, GIGI recognizes that this inheritance vectors file is empty but will proceed to impute genotypes. For trios, imputing genotypes using GIGI is equivalent to imputing genotypes conditional on the pedigree structure and minor allele frequencies.

Information about linkage-disequilibrium is potentially useful for genotype imputation on trios. Since the current version of GIGI does not leverage information from linkage-disequilibrium, you may want to use existing linkage-disequilibrium-based software that supports imputation on trios (e.g. BEAGLE).