Premium
Using linkage analysis to identify novel gene‐gene interactions in Alzheimer’s disease
Author(s) -
Grunin Michelle,
Wheeler Nicholas R.,
Bush William S.,
Haines Jonathan L.
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.043435
Subject(s) - genetics , biology , locus (genetics) , gene family , gene , genetic linkage , genome , allele , computational biology , multiplex
Background Linkage analysis was a workhorse of disease gene mapping until the advent of more refined genomic scans through genome wide association studies (GWAS). However, GWAS’ ability to detect gene‐gene interactions is hampered by the required huge sample sizes. Whole Genome Sequencing (WGS) captures nearly all allelic variation enabling direct measurement of co‐segregation with disease for each allele in a family through linkage analysis without the need to consider recombination fractions. We developed an approach integrating co‐segregation data with extant biological interaction data to identify interacting genes involved in disease using WGS of multiplex Alzheimer’s Disease (AD) families. Method Two point linkage (MERLIN) was performed on WGS of 46 AD Sequencing Project (ADSP) families. Per family, an empirical maximum LOD score (MaxLOD) was calculated. Interactions between proteins were identified by querying multiple functional/pathway databases. Variants reaching a family‐specific MaxLOD were filtered by high/moderate impact (missense+stop/loss) and functional interaction. Further families from the ADSP are being analyzed using the same pipeline. Result Genes of interest were defined as genes of known interaction where at least 2 biologically interacting loci per family reached a family‐specific MaxLOD and were found in >1 family, creating a single shared transchromosomal locus. The transchromosomal MaxLODs were combined across multiple families sharing the same biologically interacting loci. Multiple families had biologically interacting genes with individual variants attaining a MaxLOD>1.2. One interaction of shared gene pairs/family had a transchromosomal MaxLOD across 3 families of 4.05, linking variants in SLC10A6, DHCR7, SOAT1, ABCA1 and SREBF2 in the cholesterol metabolism and lipid processing pathways. An additional 5 families had MaxLOD variants in at least one of these genes. 4 families had interacting pairs with a transchromosomal MaxLOD score >2.4 unique to the family. Conclusion This approach utilizes biological interactions to identify potential disease loci despite genomic distance, thus identifying possible epistatic effects not easily observable in case‐control datasets. It is easily applicable to any complex disorder having multiplex families.