The Alzheimer's Disease Neuroimaging Initiative,

bioRxiv2017

A common haplotype lowers SPI1 (PU.1) expression in myeloid cells and delays age at onset for Alzheimer’s disease

In this study we used age at onset of Alzheimers disease (AD), cerebrospinal fluid (CSF) biomarkers, and cis-expression quantitative trait loci (cis-eQTL) datasets to identify candidate causal genes and mechanisms underlying AD GWAS loci. In a genome-wide survival analysis of 40,255 samples, eight of the previously reported AD risk loci are significantly (P < 5x10-8) or suggestively (P < 1x10-5) associated with age at onset-defined survival (AAOS) and a further fourteen novel loci reached suggestive significance. Using stratified LD score regression we demonstrated a significant enrichment of AD heritability in hematopoietic cells of the myeloid and B-lymphoid lineage. We then investigated the impact of these 22 AAOS-associated variants on CSF biomarkers and gene expression in cells of the myeloid lineage. In particular, the minor allele of rs1057233 (G), within the previously reported CELF1 AD risk locus, shows association with higher age at onset of AD (P=8.40x10-6), higher CSF levels of A{beta}42 (P=1.2x10-4), and lower expression of SPI1 in monocytes (P=1.50x10-105) and macrophages (P=6.41x10-87). SPI1 encodes PU.1, a transcription factor critical for myeloid cell development and function. AD heritability is enriched within the SPI1 cistromes of monocytes and macrophages, implicating a myeloid PU.1 target gene network in the etiology of AD. Finally, experimentally altered PU.1 levels are correlated with phagocytic activity of BV2 mouse microglial cells and specific changes in the expression of multiple myeloid-expressed genes, including the mouse orthologs of AD-associated genes, APOE, CLU/APOJ, CD33, MS4A4A/MS4A6A, and TYROBP. Our results collectively suggest that lower SPI1 expression reduces AD risk by modulating myeloid cell gene expression and function.

neuroscience↗

bioRxiv2016

High-dimensional regression over disease subgroups

We consider high-dimensional regression over subgroups of observations. Our work is motivated by biomedical problems, where disease subtypes, for example, may differ with respect to underlying regression models, but sample sizes at the subgroup-level may be limited. We focus on the case in which subgroup-specific models may be expected to be similar but not necessarily identical. Our approach is to treat subgroups as related problem instances and jointly estimate subgroup-specific regression coefficients. This is done in a penalized framework, combining an{ell} 1 term with an additional term that penalizes differences between subgroup-specific coefficients. This gives solutions that are globally sparse but that allow information-sharing between the subgroups. We present algorithms for estimation and empirical results on simulated data and using Alzheimers disease, amyotrophic lateral sclerosis and cancer datasets. These examples demonstrate the gains our approach can offer in terms of prediction and the ability to estimate subgroup-specific sparsity patterns.

bioinformatics↗