|
Activity Number:
|
288
- SLDS CSpeed 5
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 11, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Statistical Learning and Data Science
|
|
Abstract #318401
|
|
|
Title:
|
Unsupervised Clustering of Aging Individuals Using Multi-Region Brain Transcriptomes
|
|
Author(s):
|
Annie J Lee* and Yiyi Ma and Lei Yu and Robert J. Dawe and Konstantinos Arfanakis and Richard Mayeux and Bennett David and Hans-Ulrich Klein and Philip L. De Jager
|
|
Companies:
|
Columbia University and Columbia University and Rush Alzheimer Disease Center, Rush University Medical Center and Rush Alzheimer Disease Center, Rush University Medical Center and Rush Alzheimer Disease Center, Rush University Medical Center and Columbia University and Rush Alzheimer Disease Center, Rush University Medical Center and Columbia University and Columbia University
|
|
Keywords:
|
canonical correlation analysis;
k-means clustering;
non-negative matrix factorization;
meta-analysis
|
|
Abstract:
|
Although identification of the structure of older population has not been extensively explored, the unsupervised clustering of individuals using multi-region brain transcriptomes can uncover molecular subtypes of aging individuals that have clinical and pathological implications. We preprocessed transcriptomic RNA sequencing profiles from three brain regions and deployed the canonical correlation analysis in the discovery set of individuals that have transcription profiles in all three regions. The subgroups of participants were identified through unsupervised k-means clustering for each region and then integrated into “meta-clusters” through a non-negative matrix factorization. The clustering analysis was replicated in the remaining set of individuals that have data in only one or two of the three regions. To assess the clinicopathologic relevance of the meta-clusters, we assessed whether clinical and pathologic features are different between the meta-clusters by conducting a meta-analysis of discovery and replication cohorts. We further investigated the mechanisms underlying the differences in clinical phenotypes between the two groups. The clustering analysis revealed two stable
|
Authors who are presenting talks have a * after their name.
