Biology subjects

Gaiteri, C.

Publications and source records attributed to Gaiteri, C..

At least 19 recordsLinked to original sources

bioRxiv2026

A coordinated and reproducible multicellular response to AD pathology

While single-cell measurements of the Alzheimer's disease (AD) brain allow investigation of vulnerable cells in unprecedented depth, reported findings are heterogeneous, mostly descriptive and often ignore interdependence of cellular responses. Operationalization of these findings requires a multi-cellular model of AD that is robust and reproducible. Here, we utilize a coordinated multicellular community approach across 2.3 million nuclei from 427 ROSMAP donors to identify robust multicellular axes associated with AD pathology and sex. The AD signature comprised profound alteration to RORB-GABRG excitatory neurons and disrupted oligodendrocyte precursor cell signaling. A separate set of AD-related cell-type specific molecular processes with strong sexual dimorphism also emerged. Importantly, these signatures developed and ROSMAP predicted neuropathology and sex, respectively, in a completely independent dataset (SEA-AD). These findings altogether summarize a highly consistent multi-cellular response to AD pathology in the human brain, creating a baseline cellular model to which perturbation and treatment experiments can be compared.

neuroscience↗

bioRxiv2026

Disease relevance and replicability of deep learning gene expression prediction

Recent deep learning (DL) models predict average gene expression levels from DNA sequences with high overall correlation to measured values. We examine these DL models through the lens of disease research. The seemingly high overall performance of DL models is largely due to capturing whether genes are "On" or "Off", and to a lesser extent, disease-relevant expression level changes for "On" genes. Indeed, the more bimodal the gene expression distribution, the better the reported performance. We track the extent of this issue across tissues, molecular systems, and cancers. Compounding the model evaluation issue, we find inconsistencies between the published code and reported performance, highlighting the importance of versioning and publishing performance evaluation code. These findings indicate that the high reported performance of popular DL models falls unexpectedly short in disease applications, and that the problem of personalized genomic prediction remains far from solved in a disease context.

genomics↗

bioRxiv2026

A Multi-Omics Study of Cognitive Resilience to Alzheimer's Disease

Alzheimers disease (AD) is the most severe form of dementia. While significant efforts have been made to identify AD risk factors and develop therapeutics, much less attention has been given to neurobiological factors and molecular signatures associated with cognitive resilience. Substantial evidence suggests that modifiable factors, such as physical and mental activity, may contribute to cognitive reserve and resilience. To this end, we developed a resilient cohort of paired genetics, transcriptomics and proteomics data from the prefrontal cortex of 282 cognitively normal older adults. We conducted extensive preprocessing and rigorous quality control (QC), including QC on sequence metrics and sample matching across multiomics. This cohort can be leveraged for identification of molecular signatures of cognitive resilience to AD and is freely shared with the research community.

genomics↗

bioRxiv2026

Multi-Stage Graph Attention Networks for Interpretable Alzheimer's Disease Classification from Genome-Wide Association Data

BackgroundGenome-wide association studies have identified numerous variants associated with neuropsychiatric disorders. Although some significant loci can carry substantial risk, as in Alzheimers Disease, the remaining genetic variance is distributed across many small-effect loci. Polygenic risk scores (PRS) aggregate this risk but do not capture epistatic interactions, and offer limited biological interpretability and predictive accuracy. Computing gene level risk scores and integrating known or statistically validated gene-gene associations has the potential to increase interpretability and/or accuracy. Graph Neural Networks (GNNs) can leverage graph structured genetic data that models potential epistatic interactions to achieve these goals. MethodsWe developed a three-stage Graph Attention Network (GAT) classifier using individual-level GWAS data from 7,358 participants across seven Alzheimers Disease Center cohorts. Nodes were defined as genes, with risk scores from AD and 11 genetically correlated phenotypes serving as features. We evaluated two graph construction strategies: gene co-expression networks derived from hippocampal transcriptomic data and curated pathway-based graphs. Additionally, a bilinear context module was incorporated to capture global gene-gene interactions beyond the graph topology. In Stage 1, a GNN encoder was trained on the graphs; Stage 2 injected PRS for non-coding SNPs after the encoder to better capture genetic risk via transfer learning, and Stage 3 applied adversarial training with gradient reversal for ancestry debiasing. GNN predictions were ensembled with whole-genome PRS using elastic net regression. ResultsThe best-performing GNN model -- a GAT with bilinear context operating on the pathway graph -- achieved an AUROC of 0.78 (95% CI: 0.75-0.80). Ensemble models combining Stage 2 or 3 GNN logits with whole-genome PRS achieved an AUROC of 0.82 (0.79-0.84), outperforming PRS alone (0.80). GxI attribution and additional explainability analyses revealed stage-specific biological signals, some of which re-capitulated known gene-phenotype associations and others which may reflect potential new areas of inquiry. ConclusionA multi-stage GAT framework captures complementary, non-additive genetic signal that, when ensembled with PRS, improves the accuracy of AD classification. Post-hoc explainability analyses yield biologically interpretable gene networks, supporting the utility of graph-based deep learning for dissecting complex genetic architectures.

neuroscience↗

bioRxiv2025

A hidden network of gene relationships unifies behavioral and molecular research

Behavioral and molecular approaches often share topics of inquiry across basic and applied research fields. However, for historical and practical reasons, these approaches conceptualize scientific questions differently. This segregation is particularly damaging to translational fields that are in need of the combined strength of these two major approaches. While studies are increasingly recognizing the value of integrated approaches through hybrid methods, these efforts require great resources. To potentially relieve this interdisciplinary detachment, we uncover similarities between behavioral and molecular work in the publication network to test if the two approaches are more aligned than they appear, through deeper common concepts. To do this, we first create a model to predict genes that are relevant to behavioral publications from their abstracts. These predictions are in turn used to define gene-based similarity across the entire medical literature network. This model is built on-top of a Large Language Model (LLM), used to find patterns in the semantic structure of abstracts. Our experiments demonstrated the models efficacy in predicting gene annotations in molecular abstracts and showed the practical value of predictions made on behavioral research. We found similar increased relative risks for differentially expressed genes in disease-specific publications compared to the global publication population in both the molecular and behavioral groups. Additionally, the model was able to correlate gene predictions between molecular publications and their behavioral references. We instantiated this model in a free and easy to use search engine that finds related publications via the latent representations weve established. Using this model, Abstract2Gene facilitates novel collaborations by uncovering shared gene signatures across medical literature, broadening the reach of research to distant fields and promoting collaboration between molecular and behavioral researchers.

systems biology↗

bioRxiv2025

Integration of aged brain multi-omics reveals cross-system mechanisms underlying Alzheimer's disease heterogeneity

The molecular correlates of Alzheimers disease (AD) are increasingly being defined by omics. Yet, the findings from different data types or cohorts are often difficult to reconcile. Collecting multiple omics from the same individuals allows a comprehensive view of disease-related molecular mechanisms, while addressing conflicting findings derived from single omics. Such same-sample multi-omics can reveal, for instance, when changes observed in the transcriptome share distinct but coordinated signals in epigenetics and proteomics, relationships otherwise unclear. Here, we apply a data-driven multi-omic framework to integrate epigenomic, transcriptomic, proteomic, metabolomic, and cell-type-specific population data from up to 1,358 aged human brain samples from the Religious Orders Study (ROS) and Rush Memory and Aging Project (MAP). We demonstrate the existence of sprawling cross-omics cross-system biological factors that also relate to AD phenotypes. The strongest AD-associated factor (factor 8) involved elevated immune activity at the epigenetic level, decreased expression of heat shock genes in the transcriptome, and disrupted energy metabolism and cytoskeletal dynamics in the proteome. We also showed immune-related factors (factors 2 and 3) with discordant enrichments, reflecting reactive-like glial subpopulations and protective contributions from surveillance microglia. Both were negatively associated with AD pathology, suggesting potential immune resilience mechanisms. Finally, unsupervised clustering of participants revealed eleven molecular subtypes of the aging brain, including three clusters strongly associated with AD but displaying distinct molecular signatures and phenotypic characteristics. Our findings provide a comprehensive map of molecular mechanisms underlying AD heterogeneity, highlighting the complex role of neuroinflammatory processes, and yielding potential novel biomarkers and therapeutic targets for precision medicine approaches to AD treatment.

molecular biology↗

bioRxiv2025

Gene module-trait network analysis uncovers cell type specific systems and genes relevant to Alzheimers Disease

Alzheimers Disease (AD) is marked by the accumulation of pathology, neuronal loss, and gliosis and frequently accompanied by cognitive decline. Understanding brain cell interactions is key to identifying new therapeutic targets to slow its progression. Here, we used systems biology methods to analyze single-nucleus RNA sequencing (snRNASeq) data generated from dorsolateral prefrontal cortex (DLPFC) tissues of 424 participants in the Religious Orders Study or the Rush Memory and Aging Project (ROSMAP). We identified modules of co-regulated genes in seven major cell types, assigned them to coherent cellular processes, and assessed which modules were associated with AD traits such as cognitive decline, tangle density, and amyloid-{beta} deposition. Coexpression network structure was conserved in the majority of modules across cell types, but we also found distinct communities with altered connectivity, especially when compared to bulk RNASeq, suggesting cell-specific gene co-regulation. These coexpression modules can also capture signatures of cell subpopulations and be influenced by cell proportions. Using a Bayesian network framework, we modeled the direction of relationships between the modules and AD progression. We highlight two key modules, a microglia module (mic_M46), associated with tangles; and an astrocyte module (ast_M19), associated with cognitive decline. Our work provides cell-specific molecular networks modeling the molecular events leading to AD.

systems biology↗

bioRxiv2024

Spatial Expression of Long Non-Coding RNAs in Human Brains of Alzheimer's Disease

BackgroundLong non-coding RNAs (lncRNAs) are critical regulators of physiological and pathological processes, with their dysregulation increasingly implicated in aging and Alzheimers disease (AD). To investigate the spatial and cellular distribution of lncRNAs in the aging brain, we leveraged published spatial transcriptomics (ST), single-nucleus RNA sequencing (snRNA-seq), and bulk RNA-seq datasets from the dorsolateral prefrontal cortex (DLPFC) of ROSMAP participants with and without pathological AD. ResultsLncRNAs exhibited greater subregion-specific expression than mRNAs, with enrichment in antisense and lincRNA biotypes. Subregion-enriched lncRNAs were generally not cell-type specific, and vice versa. Differential expression analysis of ST data identified AD-associated lncRNAs with distinct spatial patterns and moderate overlap with differentially expressed (DE) lncRNAs from bulk RNA-seq. Gene set enrichment revealed their involvement in chromatin remodeling, epigenetic regulation, and RNA metabolism. We also identified AD DE lncRNAs across major brain cell types using snRNA-seq but overlap with ST DE lncRNAs was limited. Among previously reported lncRNAs, OIP5-AS1 was consistently upregulated in AD in all cortical subregions. Antisense oligonucleotide (ASO) knockdown of OIP5-AS1 in iPSC-derived microglia led to upregulation of pro-inflammatory genes and downregulation of DNA replication and repair pathways. Immunoassays confirmed increased secretion of pro-inflammatory cytokines. The knockdown expression pattern was enriched for microglia-specific AD DE genes and microglia states. ConclusionsThis study provides a spatial and cellular map of lncRNAs in the aging human cortex and identifies subregion-and cell-type-enriched DE lncRNAs in AD. Our findings implicate OIP5-AS1 in microglial activation, suggesting its potential contribution to AD pathogenesis.

neuroscience↗

bioRxiv2024

The YTHDF Proteins Shape the Brain Gene Signatures of Alzheimer's Disease

The gene signatures of Alzheimers Disease (AD) brains reflect an output of a complex interplay of genetic, epigenetic, epi-transcriptomic, and post-transcriptional regulation., yet the dominant factor shaping these signatures remains unclear. To identify the most significant factor that shapes the AD brain signatures, we integrated cellular and molecular features with differential gene expression in in an explainable machine learning framework. Our result indicates that YTHDF proteins, the canonical readers of N6-methyladenosine RNA modification (m6A), are the most influential predictors of the AD brain signatures. We then show that protein modules containing YTHDFs are downregulated in human AD brains, and knocking down and pharmacologically inhibiting YTHDFs in iPSC-derived 2D and 3D neuronal models recapitulate the AD-associated transcriptional signatures. Furthermore, eCLIP-seq analysis revealed that YTHDF proteins influence AD signatures through both m6A-dependent and independent pathways. These results highlight the central role of YTHDF proteins in shaping the gene signatures of AD brains.

neuroscience↗

bioRxiv2024

Uncovering Plaque-Glia Niches in Human Alzheimer's Disease Brains Using Spatial Transcriptomics

Amyloid-beta (A{beta}) plaques and surrounding glial activation are prominent histopathological hallmarks of Alzheimers Disease (AD). However, it is unclear how A{beta} plaques interact with surrounding glial cells in the human brain. Here, we applied spatial transcriptomics (ST) and immunohistochemistry (IHC) for A{beta}, GFAP, and IBA1 to acquire data from 258,987 ST spots within 78 postmortem brain sections of 21 individuals. By coupling ST and adjacent-section IHC, we showed that low A{beta} spots exhibit transcriptomic profiles indicative of greater neuronal loss than high A{beta} spots, and high-glia spots present transcriptomic changes indicative of more significant inflammation and neurodegeneration. Furthermore, we observed that this ST glial response bears signatures of reported mouse gene modules of plaque-induced genes (PIG), oligodendrocyte (OLIG) response, disease-associated microglia (DAM), and disease-associated astrocytes (DAA), as well as different microglia (MG) states identified in human AD brains, indicating that multiple glial cell states arise around plaques and contribute to local immune response. We then validated the observed effects of A{beta} on cell apoptosis and plaque-surrounding glia on inflammation and synaptic loss using IHC. In addition, transcriptomic changes of iPSC-derived microglia-like cells upon short-interval A{beta} treatment mimic the ST glial response and mirror the reported activated MG states. Our results demonstrate an exacerbation of synaptic and neuronal loss in low-A{beta} or high-glia areas, indicating that microglia response to A{beta}-oligomers likely initiates glial activation in plaque-glia niches. Our study lays the groundwork for future pathology genomics studies, opening the door for investigating pathological heterogeneity and causal effects in neurodegenerative diseases.

genomics↗

bioRxiv2024

Human brain aging heterogeneity observed from multi-region omics data reveals a subtype closely related to Alzheimer's disease

INTRODUCTION: The interconnection between brain aging and Alzheimers disease (AD) remain to be elucidated. METHODS: We investigated multi-omics (transcriptomics and proteomics) data from multiple brain regions (i.e., the hippocampus (HIPP), prefrontal cortex (PFC), and cerebellum (CRBL)) in cognitively normal individuals. RESULTS: We found that brain samples could be divided into ADL (AD-like) and NL (normal) subtypes which were correlated across brain regions. The differentially expressed genes in the ADL samples highly overlapped with AD gene signatures and the changes were consistent across brain regions (PFC and HIPP) in the multi-omics data. Intriguingly, the ADL subtype in PFC showed more differentially expressed genes than other brain regions, which could be explained by the baseline gene expression differences in the PFC NL samples. DISCUSSION: We conclude that brain aging heterogeneity widely exists, and our findings corroborate with the hypothesis that AD-related changes occur decades before the clinical manifestation of cognitive impairment in a sub-population.

neuroscience↗

bioRxiv2023

A Molecular Basis of Human Brain Connectivity

Neuroimaging is commonly used to infer human brain connectivity, but those measurements are far-removed from the molecular underpinnings at synapses. To uncover the molecular basis of human brain connectivity, we analyzed a unique cohort of 98 individuals who provided neuroimaging and genetic data contemporaneous with dendritic spine morphometric, proteomic, and gene expression data from the superior frontal and inferior temporal gyri. Through cellular contextualization of the molecular data with dendritic spine morphology, we identified hundreds of proteins related to synapses, energy metabolism, and RNA processing that explain between-individual differences in functional connectivity and structural covariation. By integrating data at the genetic, molecular, subcellular, and tissue levels, we bridged the divergent fields of molecular biology and neuroimaging to identify a molecular basis of brain connectivity. One-Sentence SummaryDendritic spine morphometry and synaptic proteins unite the divergent fields of molecular biology and neuroimaging.

neuroscience↗

bioRxiv2023

How far are we from personalized gene expression prediction using sequence-to-expression deep neural networks?

Introductory Paragraph Deep learning methods have recently become the state-of-the-art in a variety of regulatory genomic tasks1-6 including the prediction of gene expression from genomic DNA. As such, these methods promise to serve as important tools in interpreting the full spectrum of genetic variation observed in personal genomes. Previous evaluation strategies have assessed their predictions of gene expression across genomic regions, however, systematic benchmarking is lacking to assess their predictions across individuals, which would directly evaluates their utility as personal DNA interpreters. We used paired Whole Genome Sequencing and gene expression from 839 individuals in the ROSMAP study7 to evaluate the ability of current methods to predict gene expression variation across individuals at varied loci. Our approach identifies a limitation of current methods to correctly predict the direction of variant effects. We show that this limitation stems from insufficiently learnt sequence motif grammar, and suggest new model training strategies to improve performance.

bioinformatics↗

bioRxiv2022

Dissecting the Human Leptomeninges at single-cell resolution

Emerging evidence shows that the meninges conduct essential immune surveillance and immune defense at the brain border, and the dysfunction of meningeal immunity contributes to aging and neurodegeneration. However, no study exists on the molecular properties of cell types within human leptomeninges. Here, we provide the first single nuclei profiling of dissected postmortem leptomeninges from aged individuals. We detect diverse cell types, including unique meningeal endothelial, mural, and fibroblast subtypes. For immune cells, we show that most T cells express CD8 and bear characteristics of tissue-resident memory T cells. We also identify distinct subtypes of border-associated macrophages (BAMs) that display differential gene expressions from microglia and express risk genes for Alzheimers Disease (AD), as nominated by genome-wide association studies (GWAS). We discover cell-type-specific differentially expressed genes in individuals with Alzheimers dementia, particularly in fibroblasts and BAMs. Indeed, when cultured, leptomeningeal cells display the signature of ex vivo AD fibroblasts upon amyloid-{beta} treatment. We further explore ligand-receptor interactions within the leptomeningeal niche and computationally infer intercellular communications in AD. Thus, our study establishes a molecular map of human leptomeningeal cell types, providing significant insight into the border immune and fibrotic responses in AD.

genomics↗

bioRxiv2022

The molecular genetic landscape of human brain size variation

Human brain size increases dynamically through early development, peaks in adolescence, and varies up to two-fold among adults. Although previous studies have elucidated changes in brain size across evolution, development, traits, and diseases, the molecular underpinnings of interindividual variation in brain size remain unknown. Here, we leverage postmortem brain RNA sequencing and estimates of brain weight (BW) in 2,531 individuals across three independent datasets, to identify 928 genes that show genome-wide significant associations with either higher or lower BW (BW+, BW-, respectively). These BW gene sets showed distinct neurodevelopmental trajectories and spatial patterns that mapped onto developmental, functional and cellular axes of brain organization. Expression differences among evolutionarily conserved BW genes were predictive of interspecies differences in brain size, and functional annotation of BW genes revealed enrichment for neurogenesis and cell-cell communication. Genome-wide, transcriptome-wide, and phenome-wide association analyses of in vivo neuroimaging phenotypes confirmed that the genetic regulation of BW-transcripts influences cortical surface area and volume, as well as behavioral traits related to brain function and disease. Cumulatively, our study represents a major step towards the goal of delineating the causal mechanisms of human brain size variation in health and disease.

neuroscience↗

bioRxiv2022

Aging the Brain: Multi-Region Methylation Principal Component Based Clock in the Context of Alzheimer's Disease

Alzheimers disease (AD) risk increases exponentially with age and is associated with multiple molecular hallmarks of aging, one of which is epigenetic alterations. Epigenetic age predictors based on 5 cytosine methylation (DNAm) have previously suggested that biological age acceleration may occur in AD brain tissue. To further investigate brain epigenetic aging in AD, we generated a novel age predictor termed PCBrainAge that was trained solely in cortical samples. This predictor utilizes a combination of principal components analysis and regularized regression, which reduces technical noise and greatly improves test-retest reliability. For further testing, we generated DNAm data from multiple brain regions in a sample from the Religious Orders Study and Rush Memory & Aging Project. PCBrainAge captures meaningful heterogeneity of aging, calculated according to an individuals age acceleration beyond expectation. Its acceleration demonstrates stronger associations with clinical AD dementia, pathologic AD, and APOE {varepsilon}4 carrier status compared to extant epigenetic age predictors. It does so across multiple cortical and subcortical regions. Overall, PCBrainAge is useful for investigating heterogeneity in brain aging, as well as epigenetic alterations underlying AD risk and resilience.

bioinformatics↗

bioRxiv2021

Large-Scale Deep Multi-Layer Analysis of Alzheimer's Disease Brain Reveals Strong Proteomic Disease-Related Changes Not Observed at the RNA Level

The biological processes that are disrupted in the Alzheimers disease (AD) brain remain incompletely understood. We recently performed a proteomic analysis of >2000 brains to better understand these changes, which highlighted alterations in astrocytes and microglia as likely key drivers of disease. Here, we extend this analysis by analyzing >1000 brain tissues using a tandem mass tag mass spectrometry (TMT-MS) pipeline, which allowed us to nearly triple the number of quantified proteins across cases. A consensus protein co-expression network analysis of this deeper dataset revealed new co-expression modules that were highly preserved across cohorts and brain regions, and strongly altered in AD. Nearly half of the protein co-expression modules, including modules significantly altered in AD, were not observed in RNA networks from the same cohorts and brain regions, highlighting the proteopathic nature of AD. Two such AD-associated modules unique to the proteomic network included a module related to MAPK signaling and metabolism, and a module related to the matrisome. Analysis of paired genomic and proteomic data within subjects showed that expression level of the matrisome module was influenced by the APOE {varepsilon}4 allele, but was not related to the rate of cognitive decline after adjustment for neuropathology. In contrast, the MAPK/metabolism module was strongly associated with the rate of cognitive decline. Disease-associated modules unique to the proteome are sources of promising therapeutic targets and biomarkers for AD.

systems biology↗

bioRxiv2020

Decoding differential gene expression

Identifying the molecular mechanisms that control differential gene expression (DE) is a major goal of basic and disease biology. Combining the strengths of systems biology and deep learning in a model called DEcode, we are able to predict DE more accurately than traditional sequence-based methods, which do not utilize systems biology data. To determine the biological origins of this accuracy, we identify the most predictive regulators and types of regulatory interactions in DEcode, contrasting their roles across many human tissues. Diverse systems biology, ontological and disease-related assessments all point to the predominant influence of post-translational RNA-binding factors on DE. Through the combinatorial gene regulation that is captured in DEcode, it is even possible to predict relatively subtle person-to-person variation in gene expression. We demonstrate the broad applicability of these clinically-relevant predictions by predicting drivers of aging throughout the human lifespan, gene coexpression relationships on a genome-wide scale, and frequent DE in diverse conditions. Researchers can freely access DEcode to utilize genomic big data in identifying influential molecular mechanisms for any human expression data - www.differentialexpression.org.

systems biology↗