Biology subjects

Singleton, A. B.

Publications and source records attributed to Singleton, A. B..

At least 19 recordsLinked to original sources

bioRxiv2026

Profiling of Aggregated Amyloid β42-induced Proteomic Alterations in KOLF2.1J Induced Pluripotent Stem Cell-derived Neurons

Amyloid[beta]; (A[beta];) plaques are a hallmark of Alzheimer[prime]s disease (AD). A human cellular neuronal model that recaptures A[beta];-induced pathology is critical for advancing AD research. However, comprehensive proteomic profiling of A[beta];-induced cellular model remains elusive. In this study, we investigated the proteomic changes in induced pluripotent stem cell (iPSC)-derived neurons (iNs) exposed to synthetic A[beta]; (1-42) peptides (A[beta];42A) to improve our understanding of the cellular responses of A[beta]; aggregates and to establish a human-related platform for AD research. A[beta];42A formed extracellular aggregates around neuronal soma and neurites, impaired neurite outgrowth, and induced the expression of multiple AD-associated genes, such as APOE, BACE1, ADAM10. To define the molecular landscape of A[beta];-induced neuronal dysfunction, we performed proteomic analyses of whole-cell lysates as well as soma- and neurite-enriched fractions. Proteomic profiling revealed extensive gene alterations in pathways associated with synaptic function, neuronal maintenance, and AD pathogenesis, such as the upregulation of APOE. Importantly, the A[beta];42A-iNs system recapitulated molecular signatures observed in AD brain tissue and cerebrospinal fluid. Representative protein changes, including APOE upregulation and its colocalization with A[beta]; aggregates, were further validated in postmortem human AD brain tissue. Together, these findings manifest that the A[beta];42A-iNs system reproduces multiple cellular and molecular features of AD and exhibits strongly consensus with clinical observations. It provides a valuable platform for investigating A[beta];-driven neurodegeneration and for the discovery of therapeutic targets for AD.

neuroscience↗

bioRxiv2026

Long-read sequencing maps transposable element variation and its regulatory and epigenetic effects in the human brain

Transposable elements (TEs) are mobile DNA sequences that shape genome architecture and gene regulation, yet their roles in the human brain remain largely unresolved. Short-read sequencing lacks the resolution to accurately map TE insertions, detect associated structural variants, and resolve highly repetitive regions. Here, we leverage long-read whole-genome sequencing to profile germline TE insertions in postmortem brain tissue from two ancestrally diverse cohorts: the North American Brain Expression Consortium (NABEC; European ancestry, n = 205) and the Human Brain Collection Core (HBCC; African and African-admixed ancestry, n = 146). We identified 2,842 and 1,660 high-confidence non-reference insertions in HBCC and NABEC, respectively, spanning Alu, LINE-1, and SVA elements. We then also further characterized complex short tandem repeat and variable number tandem repeat variation within reference SVA and Alu loci. Reference TEs were also found to mediate complex structural variants at loci implicated in brain development and neurodegenerative disease, with several showing ancestry-specific patterns. Integration of bulk RNA-sequencing data identified TE expression quantitative trait loci, including insertions that modulate neuronal gene expression. Single-nucleus RNA sequencing revealed cell-type-specific effects of TE regulation across cortical populations. Long-read methylation profiling further demonstrated age-associated epigenetic regulation of both reference and non-reference Alu elements. As a community resource, we release a catalog of TE insertions, allele frequencies, and ancestry-specific distributions to enable future functional and disease-focused investigations. Together, these findings highlight the widespread regulatory and epigenetic influence of TEs in the human brain and establish long-read sequencing as a powerful approach for uncovering cell-type- and population-specific TE dynamics.

genomics↗

bioRxiv2025

Random forest model improves annotation and discovery of variants of uncertain significance in Alzheimer's and other neurological disorders

Variants of uncertain significance (VUS) are a bottleneck for genetic discovery and complicate clinical decision-making in Alzheimers disease and related neurological disorders (ADRD). We developed MoVUS: Model for Variants of Unknown Significance, a random-forest approach that integrates functional predictors to classify missense VUS. MoVUS leverages a balanced random forest model trained on dbNSFP v5.1a with high-confidence ClinVar and HGMD labels, using harmonized functional prediction rankscores. MoVUS produced confident, explainable calls, with [~]98% accuracy (AUC [~]0.998), prioritizing potentially pathogenic candidates and down-ranked likely benign variants on independent validation sets of ClinVar-only and HGMD-only variants. In our discovery analyses on ADRD-implicated variants in dbNSFP and from independent collaborator cohorts, we achieved high-confidence classifications on a majority of the unknown variants (average of 55% of discovery variants). We also had access to medical records and family trees for some variants, further validating our findings. Across held-out and external datasets, MoVUS reports high accuracy alongside confidence scores and helps prioritize actionable candidates, and reduces bias by considering multiple scores for each variant. To facilitate use, we developed a web app for users to browse across 100+ ADRD genes. MoVUS provides transparent, reproducible triage for research follow-up by pairing consensus predictors with SHAP-based visualizations and explanations.

genetics↗

bioRxiv2025

Haplotype-Resolved DNA Methylation at the APOE Locus identifies Allele-Specific Epigenetic Signatures Relevant to Alzheimer's Disease Risk

The APOE gene encodes a key lipid transport protein and plays a central role in Alzheimers disease (AD) pathogenesis. Three common APOE alleles, {varepsilon}2 (rs7412(C>T), {varepsilon}3 (reference), and {varepsilon}4 (rs429358(T>C)), arise from two coding variants in exon 4 and confer distinct AD risk profiles, with {varepsilon}4 increasing risk and {varepsilon}2 providing protection. The {varepsilon}3-linked APOE variant rs769455[T] has also been associated with elevated AD risk in individuals of African ancestry carrying both rs769455[T] and {varepsilon}4 alleles. These single nucleotide variants (SNVs) reside in a cytosine-phosphate-guanine (CpG) island, which is a region with a higher frequency of CpG sites compared to the rest of the genome. CpG sites are subject to 5-methylcytosine (5mC) methylation by DNA methyltransferases which add a methyl group to the fifth carbon on the cytosine residue of a CpG site. The presence of SNVs can disrupt this process, making these regions prime targets for differential methylation; however, allele-specific methylation patterns in APOE remain poorly resolved due to technical limitations of conventional bisulfite and methylation array based methods, including degraded DNA quality, sparse CpG coverage, and lack of haplotype phasing. Here, we leverage high-accuracy long-read sequencing data to generate haplotype-resolved methylation profiles of the APOE locus in 332 postmortem brain samples from two ancestrally different cohorts. This includes 201 individuals of European ancestry from the North American Brain Expression Consortium (NABEC), comprising 402 haplotypes (48 {varepsilon}2 and 58 {varepsilon}4 alleles), and 131 individuals of African and African admixed ancestry from the Human Brain Core Collection (HBCC), comprising 262 haplotypes (25 {varepsilon}2, 64 {varepsilon}4, and 7 rs769455 alleles). A linear regression analysis identified 18 novel differentially methylated CpG sites (DMCs) associated with APOE {varepsilon}2, {varepsilon}4, and rs769455 within a gene cluster spanning TOMM40, APOE, APOC1, and APOC4-APOC2. This represents the most comprehensive haplotype-resolved methylation study of APOE in human brain tissue to date. Our results uncover distinct allele-specific methylation signatures and demonstrate the power of long-read sequencing for resolving epigenetic variation relevant to AD risk.

genomics↗

bioRxiv2025

Dissecting the biological impact of GBA1 mutations using multi-omics in an isogenic setting

GBA1 is a risk gene for multiple neurodegenerative diseases, including Lewy Body Dementia and Parkinsons disease, and biallelic pathogenic variants in the gene result in the lysosomal storage disorder Gaucher disease. GBA1 encodes the enzyme glucocerebrosidase (GCase), and alterations in the gene result in reduced enzymatic activity, which affects lysosome function downstream. Induced pluripotent stem cells (iPSCs) are a useful tool for testing the functional consequences of gene variants in an isogenic setting. Additionally, they can be used to perform multiomic studies to explore biological effects independent of disease mechanisms. Using CRISPR-edited isogenic KOLF2.1J iPSC lines containing pathogenic GBA1 variants D409H (p.D448H), D409V (p.D448V) and GBA1 knockout line generated by the iPSC Neurodegenerative Disease Initiative (iNDI), we examined potential molecular mechanisms and downstream consequences of GCase reduction. In this study, we confirm that this isogenic series behaves as expected for loss of function variants, despite the known difficulties with GBA1 editing. We identified that there are limited overlapping results across cell types suggesting potential different downstream effects caused by GBA1 variants. Additionally, we note that RNA-based quantitation may not be the best method to characterize GCase mechanisms, but protein and metabolomic analyses may be used to evaluate differences across genotypes.

genomics↗

bioRxiv2025

Large-scale HLA immunopeptidome and interactome profiling in microglia

Microglia are immune cells of the brain and act as major antigen presenting cells. Antigen presentation involves the human leukocyte antigen (HLA) complex, which is implicated in genetic risk of multiple neurodegenerative diseases. How HLA affects the function of microglia in the context of neurodegenerative disease remains unclear. Here, we investigated the HLA epitopes and their protein interactome in human induced pluripotent stem cell (iPSC)-derived microglia-like cells (iMGLs) using systematic mass spectrometry (MS)-based immunopeptidomics, whole-cell proteomics, affinity purification, and prediction algorithms. Our results revealed the presence of almost 7,000 peptides presented by HLA class I and II within microglia. We further showed that the immunopeptidome landscapes of iPSCs, iMGLs and interferon-gamma (IFN{gamma}) stimulated iMGLs are all readily distinguishable. Furthermore, HLA interacts with different groups of proteins in iPSCs compared to iMGLs which involve proteins in immune response. Importantly, we detected 25 HLA epitopes derived from 15 genes associated with Alzheimers and related dementias such as Tau, PLD3 (Alzheimers disease), TDP-43, FUS (Frontotemporal dementia), and PARK7, VPS35 (Lewy Body dementia). We predicted 31 mutant epitopes derived from these ADRD genes that could be presented with strong interaction to HLA molecules. Along with these epitopes, we observed an enrichment of immune-related interaction proteins in microglia treated with IFN{gamma}. These results provide evidence that aggregated and mutated proteins can interact with HLA alleles and be presented on the cell surface by microglia cells. This study sheds light on the antigen presenting and adaptive immunity mechanism within the central nervous system and its possible effects on neurodegenerative diseases.

immunology↗

bioRxiv2025

Temporal dynamics of proteome and phosphorproteome during neuronal differentiation in the reference KOLF2.1J iPSC line

Induced pluripotent stem cell (iPSC)-derived neurons have emerged as a powerful model to investigate both neuronal development and neurodegenerative diseases. Although transcriptomics and imaging have been applied to characterize neuronal development signatures, comprehensive datasets of protein and post-translational modifications (PTMs) are not readily available. Here, we applied quantitative proteomics and phosphoproteomics to profile the differentiation of the KOLF2.1J iPSC line, the first reference line of the iPSC Neurodegenerative Disease Initiative (iNDI) project. We developed an automated workflow enabling high-coverage enrichment of proteins and phosphoproteins. Our results revealed molecular signatures across proteomic and phosphoproteomic landscapes during differentiation of iPSC-derived neurons. Proteomic data highlighted distinct changes in mitochondrial pathways throughout the course of differentiation, while phosphoproteomics revealed specific regulatory dynamics in GTPase signaling pathways and microtubule proteins. Additionally, phosphosite dynamics exhibited discordant trends compared to protein expression, particularly in processes related to axon functions and RNA transport. Furthermore, we mapped the kinase dynamic changes that are critical for neuronal development and maturation. We developed an interactive Web app (https://niacard.shinyapps.io/Phosphoproteome/) to visualize temporal landscape dynamics of protein and phosphosite expression. By establishing baselines of proteomic and phosphoproteomic profiles for neuronal differentiation, this dataset offers a valuable resource for future research into neuronal development and neurodegenerative diseases using this reference iPSC line. HighlightsO_LITemporal dynamics of proteome and phosphoproteome profiles in KOLF2.1J iPSC derived neurons. C_LIO_LIPhosphoproteomics highlights GTPase signaling and microtubule regulation in neuronal differentiation. C_LIO_LIKinome mapping reveals a shift in kinase activity patterns from early to late differentiation. C_LIO_LIShinyapp for visualizing the trajectory of protein and phosphosite expression during neuronal differentiation. C_LI

neuroscience↗

bioRxiv2025

Proteomic Analysis of Endemic Viral Infections in Neuronsoffers Insights into Neurodegenerative Diseases

Endemic viral infections with low pathogenicity are often overlooked due to their mild symptoms, yet they can exert long-term effects on cellular function and contribute to disease pathogenesis. While viral infections have been implicated in neurodegenerative disorders, their impact on the neuronal proteome remains poorly understood. Here, we differentiated human induced pluripotent stem cells (KOLF2.1J) into mature neurons to investigate virus-induced proteomic changes following infection with five neurotropic endemic human viruses: Herpes simplex virus 1 (HSV-1), Human coronavirus 229E (HCoV-229E), Epstein-Barr virus (EBV), Varicella-Zoster virus (VZV), and Influenza A virus (H1N1). Given that these viruses can infect adults and have the potential to cross the placental barrier, their molecular impact on neurons may be relevant across the lifespan. Using mass spectrometry-based proteomics with a customized library for simultaneous detection of human and viral proteins, we confirmed successful infections and identified virus-specific proteomic signatures. Notably, virus-induced protein expression changes converged on key neuronal pathways, including those associated with neurodegeneration. Gene co-expression network analysis identified protein modules correlated with viral proteins. Pathway enrichment analysis of these modules revealed associations with the nervous system, including pathways linked to Alzheimers and Parkinsons disease. Remarkably, several viral-induced proteomic alterations overlapped with changes observed in postmortem Alzheimers patient brains, suggesting a mechanistic connection between viral exposure and neurodegenerative disease progression. These findings provide molecular insights into how common viral infections perturb neuronal homeostasis and may contribute to neurodegenerative pathology, highlighting the need to consider endemic viruses as potential environmental risk factors in neurological disorders.

neuroscience↗

bioRxiv2025

CARDBiomedBench: A Benchmark for Evaluating Large Language Model Performance in Biomedical Research

BackgroundsBiomedical research requires sophisticated understanding and reasoning across multiple specializations. While large language models (LLMs) show promise in scientific applications, their capability to safely and accurately support complex biomedical research remains uncertain. MethodsWe present CARDBiomedBench, a novel question-and-answer benchmark for evaluating LLMs in biomedical research. For our pilot implementation, we focus on neurodegenerative diseases (NDDs), a domain requiring integration of genetic, molecular, and clinical knowledge. The benchmark combines expert-annotated question-answer (Q/A) pairs with semi-automated data augmentation, drawing from authoritative public resources including drug development data, genome-wide association studies (GWAS), and Summary-data based Mendelian Randomization (SMR) analyses. We evaluated seven private and open-source LLMs across ten biological categories and nine reasoning skills, using novel metrics to assess both response quality and safety. ResultsOur benchmark comprises over 68,000 Q/A pairs, enabling robust evaluation of LLM performance. Current state-of-the-art models show significant limitations: models like Claude-3.5-Sonnet demonstrates excessive caution (Response Quality Rate: 25% [95% CI: 25% {+/-} 1], Safety Rate: 76% {+/-} 1), while others like ChatGPT-4o exhibits both poor accuracy and unsafe behavior (Response Quality Rate: 37% {+/-} 1, Safety Rate: 31% {+/-} 1). These findings reveal fundamental gaps in LLMs ability to handle complex biomedical information. ConclusionCARDBiomedBench establishes a rigorous standard for assessing LLM capabilities in biomedical research. Our pilot evaluation in the NDD domain reveals critical limitations in current models ability to safely and accurately process complex scientific information. Future iterations will expand to other biomedical domains, supporting the development of more reliable AI systems for accelerating scientific discovery.

bioinformatics↗

bioRxiv2024

CNV-Finder: Streamlining Copy Number Variation Discovery

Copy Number Variations (CNVs) play pivotal roles in the etiology of complex diseases and are variable across diverse populations. Understanding the association between CNVs and disease susceptibility is significant in disease genetics research and often requires analysis of large sample sizes. One of the most cost-effective and scalable methods for detecting CNVs is based on normalized signal intensity values, such as Log R Ratio (LRR) and B Allele Frequency (BAF), from Illumina genotyping arrays. In this study, we present CNV-Finder, a novel pipeline integrating deep learning techniques on array data, specifically a Long Short-Term Memory (LSTM) network, to expedite the large-scale identification of CNVs within predefined genomic regions. This facilitates efficient prioritization of samples for time-consuming or costly subsequent analyses such as Multiplex Ligation-dependent Probe Amplification (MLPA), short-read, and long-read whole genome sequencing. We incorporate four genes to establish our methods--Parkin (PRKN), Leucine Rich Repeat And Ig Domain Containing 2 (LINGO2), Microtubule Associated Protein Tau (MAPT), and alpha-Synuclein (SNCA)--which may be relevant to neurological diseases such as Alzheimers disease (AD), Parkinsons disease (PD), Progressive Supranuclear Palsy (PSP), or related disorders such as essential tremor (ET). By training our models on expert-annotated samples and validating them across diverse cohorts, including those from the Global Parkinsons Genetics Program (GP2) and additional dementia-specific databases, we demonstrate the efficacy of CNV-Finder in accurately detecting deletions and duplications. Our pipeline outputs app-compatible files for visualization within CNV-Finders interactive web application. This interface enables researchers to review predictions and filter displayed samples by model prediction values, LRR range, and variant count in order to explore or confirm results. Our pipeline integrates this human feedback to enhance model performance and reduce false positive rates. Through a series of comprehensive analyses and validations using visual inspection, MLPA, short-read, and long-read sequencing data, we demonstrate the robustness and adaptability of CNV-Finder in identifying CNVs with regions of varied size, probe density, and noise. Our findings highlight the significance of contextual understanding and human expertise in enhancing the precision of CNV identification, particularly in complex genomic regions like 17q21.31. The CNV-Finder pipeline is a scalable, publicly available resource for the scientific community, available on GitHub (https://github.com/GP2code/CNV-Finder; DOI 10.5281/zenodo.14182563). CNV-Finder not only expedites accurate candidate identification but also significantly reduces the manual workload for researchers, enabling future targeted validation and downstream analyses in regions or phenotypes of interest.

bioinformatics↗

bioRxiv2024

GenoTools: An Open-Source Python Package for Efficient Genotype Data Quality Control and Analysis

GenoTools, a Python package, streamlines population genetics research by integrating ancestry estimation, quality control (QC), and genome-wide association studies (GWAS) capabilities into efficient pipelines. By tracking samples, variants, and quality-specific measures throughout fully customizable pipelines, users can easily manage genetics data for large and small studies. GenoTools "Ancestry" module renders highly accurate predictions, allowing for high-quality ancestry-specific studies, and enables custom ancestry model training and serialization, specified to the users genotyping or sequencing platform. As the genotype processing engine that powers several large initiatives, including the NIHs Center for Alzheimers and Related Dementias (CARD) and the Global Parkinsons Genetics Program (GP2). GenoTools was used to process and analyze the UK Biobank and major Alzheimers Disease (AD) and Parkinsons Disease (PD) datasets with over 400,000 genotypes from arrays and 5000 sequences and has led to novel discoveries in diverse populations. It has provided replicable ancestry predictions, implemented rigorous QC, and conducted genetic ancestry-specific GWAS to identify systematic errors or biases through a single command. GenoTools is a customizable tool that enables users to efficiently analyze and scale genotype data with reproducible and scalable ancestry, QC, and GWAS pipelines.

genomics↗

bioRxiv2021

A fully automated FAIMS-DIA proteomic pipeline for high-throughput characterization of iPSC-derived neurons

Fully automated proteomic pipelines have the potential to achieve deep coverage of cellular proteomes with high throughput and scalability. However, it is important to evaluate performance, including both reproducibility and ability to provide meaningful levels of biological insight. Here, we present an approach combining high field asymmetric waveform ion mobility spectrometer (FAIMS) interface and data independent acquisition (DIA) proteomics approach developed as part of the induced pluripotent stem cell (iPSC) Neurodegenerative Disease Initiative (iNDI), a large-scale effort to understand how inherited diseases may manifest in neuronal cells. Our FAIMS-DIA approach identified more than 8000 proteins per mass spectrometry (MS) acquisition as well as superior total identification, reproducibility, and accuracy compared to other existing DIA methods. Next, we applied this approach to perform a longitudinal proteomic profiling of the differentiation of iPSC-derived neurons from the KOLF2.1J parental line used in iNDI. This analysis demonstrated a steady increase in expression of mature cortical neuron markers over the course of neuron differentiation. We validated the performance of our proteomics pipeline by comparing it to single cell RNA-Seq datasets obtained in parallel, confirming expression of key markers and cell type annotations. An interactive webapp of this temporal data is available for aligned-UMAP visualization and data browsing (https://share.streamlit.io/anant-droid/singlecellumap). In summary, we report an extensively optimized and validated proteomic pipeline that will be suitable for large-scale studies such as iNDI.

neuroscience↗

bioRxiv2021

A CRISPRi/a platform in iPSC-derived microglia uncovers regulators of disease states

Microglia are emerging as key drivers of neurological diseases. However, we lack a systematic understanding of the underlying mechanisms. Here, we present a screening platform to systematically elucidate functional consequences of genetic perturbations in human iPSC-derived microglia. We developed an efficient eight-day protocol for the generation of microglia-like cells based on the inducible expression of six transcription factors. We established inducible CRISPR interference and activation in this system and conducted three screens targeting the "druggable genome". These screens uncovered genes controlling microglia survival, activation and phagocytosis, including neurodegeneration-associated genes. A screen with single-cell RNA sequencing as the readout revealed that these microglia adopt a spectrum of states mirroring those observed in human brains and identified regulators of these states. A disease-associated state characterized by SPP1 expression was selectively depleted by CSF1R inhibition. Thus, our platform can systematically uncover regulators of microglia states, enabling their functional characterization and therapeutic targeting.

neuroscience↗

bioRxiv2021

The Foundational data initiative for Parkinsons disease (FOUNDIN-PD): enabling efficient translation from genetic maps to mechanism

The FOUNdational Data INitiative for Parkinsons Disease (FOUNDIN-PD) is an international collaboration producing fundamental resources for Parkinsons disease (PD). FOUNDIN-PD generated a multi-layered molecular dataset in a cohort of induced pluripotent stem cell (iPSC) lines differentiated to dopaminergic (DA) neurons, a major affected cell type in PD. The lines were derived from the Parkinsons Progression Markers Initiative study including participants with PD carrying monogenic PD (SNCA) variants, variants with intermediate effects and variants identified by genome-wide association studies and unaffected individuals. We generated genetic, epigenetic, regulatory, transcriptomic, and longitudinal cellular imaging data from iPSC-derived DA neurons to understand molecular relationships between disease associated genetic variation and proximate molecular events. These data reveal that iPSC-derived DA neurons provide a valuable cellular context and foundational atlas for modelling PD genetic risk. We have integrated these data into a FOUNDIN-PD data browser (https://www.foundinpd.org) as a resource for understanding the molecular pathogenesis of PD.

genomics↗

bioRxiv2021

Multi-Modality Machine Learning Predicting Parkinsons Disease

BackgroundPersonalized medicine promises individualized disease prediction and treatment. The convergence of machine learning (ML) and available multi-modal data is key moving forward. We build upon previous work to deliver multi-modal predictions of Parkinsons Disease (PD). MethodsWe performed automated ML on multi-modal data from the Parkinsons Progression Marker Initiative (PPMI). After selecting the best performing algorithm, all PPMI data was used to tune the selected model. The model was validated in the Parkinsons Disease Biomarker Program (PDBP) dataset. Finally, networks were built to identify gene communities specific to PD. FindingsOur initial model showed an area under the curve (AUC) of 89.72% for the diagnosis of PD. The tuned model was then tested for validation on external data (PDBP, AUC 85.03%). Optimizing thresholds for classification, increased the diagnosis prediction accuracy (balanced accuracy) and other metrics. Combining data modalities outperforms the single biomarker paradigm. UPSIT was the largest contributing predictor for the classification of PD. The transcriptomic data was used to construct a network of disease-relevant transcripts. InterpretationWe have built a model using an automated ML pipeline to make improved multi-omic predictions of PD. The model developed improves disease risk prediction, a critical step for better assessment of PD risk. We constructed gene expression networks for the next generation of genomics-derived interventions. Our automated ML approach allows complex predictive models to be reproducible and accessible to the community. FundingNational Institute on Aging, National Institute of Neurological Disorders and Stroke, the Michael J. Fox Foundation, and the Global Parkinsons Genetics Program. RESEARCH IN CONTEXTO_ST_ABSEvidence before this studyC_ST_ABSPrior research into predictors of Parkinsons disease (PD) has either used basic statistical methods to make predictions across data modalities, or they have focused on a single data type or biomarker model. We have done this using an open-source automated machine learning (ML) framework on extensive multi-modal data, which we believe yields robust and reproducible results. We consider this the first true multi-modality ML study of PD risk classification. Added value of this studyWe used a variety of linear, non-linear, kernel, neural networks, and ensemble ML algorithms to generate an accurate classification of both cases and controls in independent datasets using data that is not involved in PD diagnosis itself at study recruitment. The model built in this paper significantly improves upon our previous models that used the entire training dataset in previous work1. Building on this earlier work, we showed that the PD diagnosis can be refined using improved algorithmic classification tools that may yield potential biological insights. We have taken careful consideration to develop and validate this model using public controlled-access datasets and an open-source ML framework to allow for reproducible and transparent results. Implications of all available evidenceTraining, validating, and tuning a diagnostic algorithm for PD will allow us to augment clinical diagnoses or risk assessments with less need for complex and expensive exams. Going forward, these models can be built on remote or asynchronously collected data which may be important in a growing telemedicine paradigm. More refined diagnostics will also increase clinical trial efficiency by potentially refining phenotyping and predicting onset, allowing providers to identify potential cases earlier. Early detection could lead to improved treatment response and higher efficacy. Finally, as part of our workflow, we built new networks representing communities of genes correlated in PD cases in a hypothesis-free manner, showing how new and existing genes may be connected and highlighting therapeutic opportunities.

genomics↗

bioRxiv2020

Genome-wide CRISPRi/a screens in human neurons link lysosomal failure to ferroptosis

Single-cell transcriptomics provide a systematic map of gene expression in different human cell types. The next challenge is to systematically understand cell-type specific gene function. The integration of CRISPR-based functional genomics and stem cell technology enables the scalable interrogation of gene function in differentiated human cells. Here, we present the first genomewide CRISPR interference and CRISPR activation screens in human neurons. We uncover pathways controlling neuronal response to chronic oxidative stress, which is implicated in neurodegenerative diseases. Unexpectedly, knockdown of the lysosomal protein prosaposin strongly sensitizes neurons, but not other cell types, to oxidative stress by triggering the formation of lipofuscin, a hallmark of aging, which traps iron, generating reactive oxygen species and triggering ferroptosis. We also determine transcriptomic changes in neurons following perturbation of genes linked to neurodegenerative diseases. To enable the systematic comparison of gene function across different human cell types, we establish a data commons named CRISPRbrain.

genomics↗

bioRxiv2020

The Parkinson's Disease GWAS Locus Browser

Parkinsons disease (PD) is a neurodegenerative disease with an often complex genetic component identifiable by genome-wide association studies (GWAS). The most recent large scale PD GWASes have identified more than 90 independent risk variants for PD risk and progression across 80 loci. One major challenge in current genomics is identifying the causal gene(s) and variant(s) from each GWAS locus. Here we present a GWAS locus browser application that combines data from multiple databases to aid in the prioritization of genes associated with PD GWAS loci. We included 92 independent genome-wide significant signals from multiple recent PD GWAS studies including the PD risk GWAS, age-at-onset GWAS and progression GWAS. We gathered data for all 2336 genes within 1Mb up and downstream of each variant to allow users to assess which gene(s) are most associated with the variant of interest based on a set of self-ranked criteria. Our aim is that the information contained in this browser (https://pdgenetics.shinyapps.io/GWASBrowser/) will assist the PD research community with the prioritization of genes for follow-up functional studies and as potential therapeutic targets.

genetics↗

bioRxiv2019

Predicting Alzheimer’s disease progression trajectory and clinical subtypes using machine learning

BackgroundAlzheimers disease (AD) is a common, age-related, neurodegenerative disease that impairs a persons ability to perform day-to-day activities. Diagnosing AD is challenging, especially in the early stages. Many patients still go undiagnosed, partly due to the complex heterogeneity in disease progression. This highlights a need for early prediction of the disease course to assist its treatment and tailor therapy options to the disease progression rate. Recent developments in machine learning techniques provide the potential to not only predict disease progression and trajectory of AD but also to classify the disease into different etiological subtypes. Methods and findingsThe work shown here clusters participants in distinct and multifaceted progression subgroups of AD and discusses an approach to predict the progression rate from baseline diagnosis. We observed that the myriad of clinically reported symptoms summarized in the proposed AD progression space corresponds directly with memory and cognitive measures, which are routinely used to monitor disease onset and progression. Our analysis demonstrated accurate prediction of disease progression after four years from the first 12 months of post-diagnosis clinical data (Area Under the Curve of 0.96 (95% confidence interval (CI), 0.92-1.0), 0.81 (95% CI, 0.74-0.88) and 0.98 (95% CI, 0.96-1.0) for slow, moderate and fast progression rate patients respectively). Further, we explored the long short-term memory (LSTM) neural networks to predict the trajectory of an individual patients progression. ConclusionThe machine learning techniques presented in this study may assist providers in identifying different progression rates and trajectories in the early stages of the disease, hence allowing for more efficient and personalized healthcare deliveries. With additional information about the progression rate of AD at hand, providers may further individualize the treatment plans. The predictive tests discussed in this study not only allow for early AD diagnosis but also facilitate the characterization of distinct AD subtypes relating to trajectories of disease progression. These findings are a crucial step forward for early disease detection. These models can be used to design improved clinical trials for AD research.

neuroscience↗