Interpretable deep learning framework towards understanding molecular changes in human brains with Alzheimer’s disease: implication for microglia activation and sex differences in AD

Abstract

AbstractINTRODUCTIONThe objective of this study is to characterize the molecular changes associated with AD from gene expression data of brain tissues taking an interpretable deep learning approach which has not been fully exploited.METHODSWe trained multi-layer perceptron (MLP) models for the classification of neuropathologically confirmed AD vs. controls using the transcriptomic data of three brain regions from the ROSMAP study. The whole disease spectrum was then modeled as a progressive trajectory. SHAP (SHapley Additive exPlanations) value was derived to explain model predictions and identify significant implicated genes for subsequent network analysis of key gene modules underlying AD progression. The framework was validated using two external datasets: the Mayo RNA-seq study cohort and the Mount Sinai Brain Bank study cohort.RESULTSThe MLP models achieved superior performance in classification and prediction in external datasets. SHAP explainer revealed common and specific transcriptomic signatures from different brain regions.DISCUSSIONWe identified common gene signatures in microglia and sex specific modules in neurons that are implicated in AD. This work paves the way for utilizing artificial intelligence approaches in studying AD at the molecular level.Research-in-ContextSystematic review: Postmortem brain transcriptomes have been analyzed to study the molecular changes associated with Alzheimer’s disease, usually by a direct contrast approach such as differential gene expression analysis. Nuanced gene regulations thus cannot be easily pinpointed from convoluted data such as those from bulk-tissue profiling. We applied a novel interpretable deep learning approach to dissect the RNA-seq data collected from three different brain regions of a large clinical cohort and identified significant genes for network analysis implicated for AD.Interpretation: Our models successfully predicted neuropathological and clinical traits in both internal and external validations. We corroborated known microglial biology in addition to revealing novel sex chromosome-linked gene contributing to sex dimorphism in AD.Future directions: The framework could have broader utility of interpreting multi-omics data such as those from single-cell profiling, to advance our understanding of molecular mechanism of complex human disease such as AD.HighlightsWe applied novel interpretable deep learning methods on the postmortem brain transcriptomes from three different brain regionsWe interpreted the models to identify the most important genes implicated for ADNetwork analysis corroborated known microglial biology and revealed novel sex specific transcriptional factors for neuronal loss in AD