Gene

Biology

Candidate gene

Computational biology

Single-nucleotide polymorphism

Genetics

Genome-wide association study

Political science

Genotype

Internal medicine

Nomination

Medicine

Disease

Alzheimer s & Dementia

Gene-based analysis are the reference approach to study the effect of rare variants in a particular phenotype. There are a plethora of software and tests for case-control studies. However, their evolution and adaptation for family-based studies, especially for complex incomplete families, has been slower. We have performed a practical examination of most recent and most widely used gene-based methods available for family-based study. We have examined the performance of 22 gene-based algorithms (including collapsing, variance-component and transmission disequilibrium tests) across eight different software – FSKAT, GSKAT, SKAT, EPCATS, FarVAT, RVGDT, Rare-IBD, PedGene - in a cohort of 285 No Hispanic Whites (NHW) families (N=1,235) with late-onset Alzheimer disease (LOAD). We generated gene-sets that included rare variants (MAF <1%, according to ExAC-non Finninsh European database), with predicted moderate or high impact, according to SnpEff. The same gene-sets were used across all 22 tests after correcting for sex and two principal components. The CALPHA with BLUP correction from FarVAT was the test that provided results closer to the expected (λ=1.26). We found the CHRD gene to be genome–wide significant (P≤7.37×10−7) by four different test. Five other genes (CLCN2, HDLBP, CPAMD8, NLRP9, MAS1L) and a former GWAS candidate gene for sporadic AD (PTK2B) presented suggestive genome-wide p-values (P≤1×10−4) by at least five tests. We found rare variants in CHRD, CLCN2 and PTK2B to be more frequent in controls and presented higher brain expression levels in controls. On the contrary, rare variants in CPAMD8 were more frequent in cases than in controls, presented perfect segregation with disease status in eight families and its expression levels were higher in cases. After a thorough review of gene-based methods for family-based studies, we propose a methodological approach to identify, with high confidence, genes associated with the studied phenotype. Collapsing tests, in particular those by FarVAT, provided the best results. We have identified novel candidate genes for familial LOAD some of which have reported interactions with known AD genes. We are currently conducting replication analysis in independent genomic datasets and exploring their involvement in AD pathways through the study of transcriptomic datasets.

P2‐105: NOMINATION OF NOVEL CANDIDATE GENES FOR FAMILIAL LATE ONSET ALZHEIMER DISEASE AFTER EVALUATION OF GENE‐BASED FAMILY‐BASED METHODS