Handling linkage disequilibrium in qualitative trait linkage analysis using dense SNPs: a two-step strategy

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Linkage disequilibrium (LD) can inflate identity-by-descent (IBD) sharing estimates in affected sibling pair (ASP) analysis when parents are ungenotyped.
• This inflation can lead to spurious evidence of linkage, even when no true linkage exists between markers and disease loci.

Purpose of the Study:

• To theoretically evaluate how inflated IBD probabilities overestimate nonparametric linkage (NPL) statistics.
• To propose and systematically evaluate a two-step strategy for handling LD in dense marker data for ASP studies.

Main Methods:

• Theoretical evaluation of IBD inflation effects on NPL statistics under linkage equilibrium.
• Implementation of a two-step strategy: Step 1 filters dense markers; Step 2 applies methods like marker thinning (MT), recursive elimination (RE), SNPLINK, and MERLIN's LD modeling.
• Performance evaluation through simulations.

Main Results:

• LOD score inflation was observed exclusively when parents were ungenotyped.
• All evaluated LD handling approaches performed similarly at given thresholds, but only RE eliminated LOD score bias.
• Simulations showed MT reduced LOD score inflation by 75% to 100% while preserving information content with an r2 threshold above 0.3 (2 SNPs/cM).

Conclusions:

• A theoretical foundation was established for how inflated IBD information from dense markers leads to NPL statistic overestimation.
• The proposed two-step processing strategy provides a robust framework for assessing methods to manage LD in genetic analyses.