This is a test version of Biostars. For the public version, visit https://www.biostars.org.
General Imputation Softwares To Impute Missing Genotypes In Candidate Gene Studies

I know that we can impute missing genotypes in GWAS studies by inferring from the Hapmap or 1000-genomes genotypes. However, candidate gene studies can not use this method. Can I use the general imputation methods (e.g Amelia or other general imputation packages in R) to impute five datasets and then combine them in the analysis? Thank you.

imputation genetics
• 4,780 views • link
updated by Genotepes Genotepes • written by guhjy1 • guhjy1

1 answer

Could you share more details about your design? Otherwise it is not easy to comment/help. Basically you can impute whatever you like if you have a panel (for instance here a set of individuals sequenced for you candidate gene.

Moreover, Amelia is not imputation program, this is a program that uses imputed (and analyzed) genotypes (SNPTEST format as far as I remember).

Again, if you describe the design, we could help

• 0 views • link
written by Genotepes Genotepes

What I meant by the R package "amelia" was Amelia II (http://gking.harvard.edu/amelia/) which is a general imputation software instead of the other Ameila (http://www.sanger.ac.uk/resources/software/amelia/). The general imputation softwares usually impute missing values by the other covariates so that I wonder if they can also be used to impute missing genotypes (from non-genetic covariates instead of other public genotypes such as the Hapmap data). I have a cross-sectional data of 800 persons and among them 500 persons have disease A whereas 300 others have no disease A. I have genotyped a common literature-cited SNP for a candidate gene and 37% of those with disease A have missing genotypes and 24% of those without disease A have missing genotypes. But most of them have many covariates. Should I listwisely delete those persons with missing genotypes or impute the missing genotypes by the general imputation softwares, e.g. http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3217865/?

• 0 views • link
written by guhjy1 • guhjy1

Log in to answer this question.

More posts like this