Premium
Integration of large‐scale molecular networks and exomic data can identify Alzheimer's disease genes
Author(s) -
Lagiesetty Yashwanth,
Bourquard Thomas,
Lichtarge Olivier
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.041965
Subject(s) - gene , biology , exome sequencing , genetics , computational biology , exome , alzheimer's disease neuroimaging initiative , disease , phenotype , alzheimer's disease , medicine , pathology
Background We introduce a new approach to identify gene variants linked to common complex diseases. This method combines whole exome sequencing and large‐scale molecular network analysis. For proof of principle, we studied the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the Alzheimer’s Disease Sequencing Project (ADSP) cohorts. In each, we evaluated the coding variants of 16,087 genes in the context of their ∼6,471,000 gene‐gene interactions, and prioritized genes that sorted cases from controls. Method The deleterious functional effect of every coding variant was quantified based on an Evolutionary Action (EA) model for the impact of similar variants across evolutionary history. For each gene, this impact was averaged over every individual from the ADNI (or the ADSP) cohort, and this average was used to modulate the edge weights of the STRING network of protein‐protein interactions. This process was carried out in cases and separately in controls, to yield two modified copies of the original STRING network, one representing the mutational perturbation from individuals affected with Alzheimer’s and, the other, representing those found in healthy individuals. We systematically compared these two graphs in order to identify, region by region and gene by gene, which parts of the network was significantly mutationally perturbed. Result Among the 100 top‐ranked in ADSP and, separately, in ADNI, a quarter are identical. This overlap indicates a robustness surpassing Genome‐wide association studies. One such gene, for example, is TOMM40, which is known to be relevant to AD. More broadly, these top‐ranked genes also significantly overlap, are significantly enriched for, or significantly diffuse to known gold standard AD genes and AD‐related diseases and pathways. These criteria for success show the top‐ranked genes are not random and are relevant to AD. Conclusion This approach to integrate sequencing data with network information promises to be a useful new tool to elucidate disease genes in AD and other complex, common diseases.