Incorporating Non-Coding Annotations into Rare Variant Analysis.

Area of Science:

• Genetics
• Bioinformatics
• Genomics

Background:

• Collapsing methods for rare variant analysis on complex traits have shown limited success.
• Variant selection significantly impacts the success of these analyses, with conventional methods filtering by consequence.
• This study explores alternative filtering approaches using bioinformatics annotations.

Purpose of the Study:

• To investigate if filtering rare variants using bioinformatics annotations improves the success of collapsing methods for complex trait analysis.
• To compare the efficacy of annotation-based filtering (e.g., CADD) against traditional consequence-based filtering.
• To assess the potential of including non-coding variants in rare variant association studies.

Main Methods:

• Candidate gene and genome-wide analyses were performed using UK10K sequence and lipids data.
• Variants were filtered using functional annotations from Combined Annotation-Dependent Depletion (CADD).
• Results were contrasted with analyses filtering by 'nonsynonymous' and 'loss of function' consequences.

Main Results:

• Filtering by CADD allowed the inclusion of intronic variants, unlike consequence-based filtering.
• CADD filtering identified an association between ANGPTL4 and High Density Lipoproteins (P = 0.02) not found with other methods.
• Annotation-based filtering (CADD, FATHMM-MKL, DANN) detected association signals missed by consequence-based filtering, and vice versa.

Conclusions:

• Bioinformatics annotations from non-coding tools are valuable for rare variant analyses.
• Annotation-based filtering expands the scope beyond coding regions, enabling discovery of causal variants in non-coding DNA.
• This approach promises to uncover more causal variants for complex traits and elucidate their functional roles in disease.