Identification of functionally connected multi-omic biomarkers for Alzheimer's disease using modularity-constrained Lasso

The ROSMAP project: aging and neurodegenerative diseases through omic sciences

The Religious Order Study and Memory and Aging Project (ROSMAP) is an initiative that integrates two longitudinal cohort studies, which have been collecting clinicopathological and molecular data since the early 1990s. This extensive dataset includes a wide array of omic data, revealing the complex interactions between molecular levels in neurodegenerative diseases (ND) and aging. Neurodegenerative diseases (ND) are frequently associated with morbidity and cognitive decline in older adults. Omics research, in conjunction with clinical variables, is crucial for advancing our understanding of the diagnosis and treatment of neurodegenerative diseases. This summary reviews the extensive omics research-encompassing genomics, transcriptomics, proteomics, metabolomics, epigenomics, and multiomics-conducted through the ROSMAP study. It highlights the significant advancements in understanding the mechanisms underlying neurodegenerative diseases, with a particular focus on Alzheimer's disease.

Upregulation of extracellular proteins in a mouse model of Alzheimer's disease

Various risk factors of Alzheimer's disease (AD) are known, such as advanced age, possession of certain genetic variants, accumulation of toxic amyloid-β (Aβ) peptides, and unhealthy lifestyle. An estimate of heritability of AD ranges from 0.13 to 0.25, indicating that its phenotypic variation is accounted for mostly by non-genetic factors. DNA methylation is regarded as an epigenetic mechanism that interfaces the genome with non-genetic factors. The Tg2576 mouse model has been insightful in AD research. These transgenic mice express a mutant form of human amyloid precursor protein linked to familial AD. At 9-13 months of age, these mice show elevated levels of Aβ peptides and cognitive impairment. The current literature lacks integrative multiomics of the animal model. We applied transcriptomics and DNA methylomics to the same brain samples from ~ 11-month-old transgenic mice. We found that genes involved in extracellular matrix structures and functions are transcriptionally upregulated, and genes involved in extracellular protein secretion and localization are differentially methylated in the transgenic mice. Integrative analysis found enrichment of GO terms related to memory and synaptic functionability. Our results indicate a possibility of transcriptional modulation by DNA methylation underlying AD neuropathology.

Pathway-based integration of multi-omics data reveals lipidomics alterations validated in an Alzheimer's disease mouse model and risk loci carriers

Alzheimer's disease (AD) is a highly prevalent neurodegenerative disorder. Despite increasing evidence of the importance of metabolic dysregulation in AD, the underlying metabolic changes that may impact amyloid plaque formation are not understood, particularly for late-onset AD. This study analyzed genome-wide association studies (GWAS), transcriptomics, and proteomics data obtained from several data repositories to obtain differentially expressed (DE) multi-omics elements in mouse models of AD. We characterized the metabolic modulation in these data sets using gene ontology, transcription factor, pathway, and cell-type enrichment analyses. A predicted lipid signature was extracted from genome-scale metabolic networks (GSMN) and subsequently validated in a lipidomic data set derived from cortical tissue of ABCA-7 null mice, a mouse model of one of the genes associated with late-onset AD. Moreover, a metabolome-wide association study (MWAS) was performed to further characterize the association between dysregulated lipid metabolism in human blood serum and genes associated with AD risk. We found 203 DE transcripts, 164 DE proteins, and 58 DE GWAS-derived mouse orthologs associated with significantly enriched metabolic biological processes. Lipid and bioenergetic metabolic pathways were significantly over-represented across the AD multi-omics data sets. Microglia and astrocytes were significantly enriched in the lipid-predominant AD-metabolic transcriptome. We also extracted a predicted lipid signature that was validated and robustly modeled class separation in the ABCA7 mice cortical lipidome, with 11 of these lipid species exhibiting statistically significant modulations. MWAS revealed 298 AD single nucleotide polymorphisms-metabolite associations, of which 70% corresponded to lipid classes. These results support the importance of lipid metabolism dysregulation in AD and highlight the suitability of mapping AD multi-omics data into GSMNs to identify metabolic alterations.

Pathway-based integration of multi-omics data reveals lipidomics alterations validated in an Alzheimer's disease mouse model and risk loci carriers

Alzheimer's disease (AD) is a highly prevalent neurodegenerative disorder. Despite increasing evidence of the importance of metabolic dysregulation in AD, the underlying metabolic changes that may impact amyloid plaque formation are not understood, particularly for late-onset AD. This study analyzed genome-wide association studies (GWAS), transcriptomics, and proteomics data obtained from several data repositories to obtain differentially expressed (DE) multi-omics elements in mouse models of AD. We characterized the metabolic modulation in these data sets using gene ontology, transcription factor, pathway, and cell-type enrichment analyses. A predicted lipid signature was extracted from genome-scale metabolic networks (GSMN) and subsequently validated in a lipidomic data set derived from cortical tissue of ABCA-7 null mice, a mouse model of one of the genes associated with late-onset AD. Moreover, a metabolome-wide association study (MWAS) was performed to further characterize the association between dysregulated lipid metabolism in human blood serum and genes associated with AD risk. We found 203 DE transcripts, 164 DE proteins, and 58 DE GWAS-derived mouse orthologs associated with significantly enriched metabolic biological processes. Lipid and bioenergetic metabolic pathways were significantly over-represented across the AD multi-omics data sets. Microglia and astrocytes were significantly enriched in the lipid-predominant AD-metabolic transcriptome. We also extracted a predicted lipid signature that was validated and robustly modeled class separation in the ABCA7 mice cortical lipidome, with 11 of these lipid species exhibiting statistically significant modulations. MWAS revealed 298 AD single nucleotide polymorphisms-metabolite associations, of which 70% corresponded to lipid classes. These results support the importance of lipid metabolism dysregulation in AD and highlight the suitability of mapping AD multi-omics data into GSMNs to identify metabolic alterations.

Robust Huber-LASSO for improved prediction of protein, metabolite and gene expression levels relying on individual genotype data

Least absolute shrinkage and selection operator (LASSO) regression is often applied to select the most promising set of single nucleotide polymorphisms (SNPs) associated with a molecular phenotype of interest. While the penalization parameter λ restricts the number of selected SNPs and the potential model overfitting, the least-squares loss function of standard LASSO regression translates into a strong dependence of statistical results on a small number of individuals with phenotypes or genotypes divergent from the majority of the study population-typically comprised of outliers and high-leverage observations. Robust methods have been developed to constrain the influence of divergent observations and generate statistical results that apply to the bulk of study data, but they have rarely been applied to genetic association studies. In this article, we review, for newcomers to the field of robust statistics, a novel version of standard LASSO that utilizes the Huber loss function. We conduct comprehensive simulations and analyze real protein, metabolite, mRNA expression and genotype data to compare the stability of penalization, the cross-iteration concordance of the model, the false-positive and true-positive rates and the prediction accuracy of standard and robust Huber-LASSO. Although the two methods showed controlled false-positive rates ≤2.1% and similar true-positive rates, robust Huber-LASSO outperformed standard LASSO in the accuracy of predicted protein, metabolite and gene expression levels using individual SNP data. The conducted simulations and real-data analyses show that robust Huber-LASSO represents a valuable alternative to standard LASSO in genetic studies of molecular phenotypes.