What are the key methodological considerations and challenges when conducting a GWAS in a consanguineous population, and what analytical approaches would you use to account for increased relatedness, autozygosity, and population structure?
I haven’t personally worked with consanguineous data, but a few things I’d be aware of are: standard HWE and excess homozygosity filters will need to be relaxed, since violations are expected and are less likely in this situation to indicate genotyping error. A GRM-based LMM GWAS should still work, but expect elevated kinship across many pairs, worth checking what effect the sparse GRM versus the full GRM have. You can also quantify autozygosity explicitly using runs of homozygosity (PLINK’s --homozyg) and see what happens if you include an F_ROH inbreeding coefficient as a covariate? Also, rare recessive variants are more likely to be homozygous in your sample, which could improve your power for recessive traits. A useful starting point is Malawski et al. (2023) would be a useful paper to read: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10580289/
Thank very much @Abdel_Abdellaoui I really appreciate your help. I hope it is possible to reach out to you in case of future GWAS in consanguineous population. Sincerely.
1 Like