DATE: June 8, 2017
TIME: 10:00AM PDT, 1:00PM EDT, 7:00PM CEST

Alzheimer’s Disease (AD) is the result of complex interactions between risk factors that cause pleiotropic changes in molecular networks linking a host of biological processes. A variety of genetic factors have been shown to contribute to risk with varying degrees of penetrance: the identification of mutations in the amyloid-beta precursor protein (APP), presenilin (PSEN1 and PSEN2) genes that cause Mendelian forms of AD represented key milestones for understanding the initial mechanisms and pathways involved in AD pathogenesis. Remarkably, variants in these genes confer a different transcriptomic profiles, and mutation carriers clustered separately from their non-carrier siblings. New evidences provide support for both neuronal and glial specific pathways contributing to pathogenesis. However, little is understood about how the genetic loci and molecular changes are organized into common networks. We combined transcriptomic cell-type profiling and network co-expression analyses to study a unique collection of human postmortem brain tissue ascertained to represent the AD Mendelian mutations.

Using novel digital deconvolution approaches, we derived cell-type specific expression. We ascertain the distribution of neuros, microglia, oligodendrocytes and astrocytes in a collection of more than 1500 AD and non-demented subjects. We derived gene regulatory networks employing the expression corrected for the distinct cell-type distributions, and identified modules that cluster genes that harbor variants usually associated with both early-onset autosomal dominant (PSEN1) and late-onset sporadic classifications of AD (SOD1, BACE1, PICALM, SLC4A2).

Understanding variant-specific effects is of an immense importance for the elucidation of the underlying biology of the Alzheimer Disease. Our initial analysis reveals a transcriptional regulation module that link that early-onset autosomal dominant and late-onset sporadic genes.

Learning Objectives:

• Identify confounding factors that can affect transcriptomic analyses and learn how to address them
• Familiarize with machine learning techniques that allow to validate results when analyzing dataset with a reduced number of samples
• Learn digital deconvolution approaches to infer cell composition from RNA-seq data
• Learn how transcriptomic profiles can reveal gene co-expression networks