Biology subjects

Prokopenko, D.

Publications and source records attributed to Prokopenko, D..

16 recordsLinked to original sources

bioRxiv2025

Analyzing associations and higher-order effects in multi-omics data with double machine learning

MotivationIntegrative omics analyses enhance our understanding of disease mechanisms and biomarkers by investigating relationships among traits, omics measurements, genetic variants, and epidemiological factors. Statistically, these analyses are challenging and require robust, flexible methodologies due to high dimensionality, non-standard data distributions, and potentially complex, non-linear confounding effects. ResultsTo facilitate the integration and analysis of multi-omics data, we introduce the Robust Omics MethodologY (ROMY) framework and its corresponding R implementation, the romy package. ROMY enables users to (a) perform robust association testing between two target variables while incorporating flexible covariate adjustments, (b) examine effects on measurement variances and covariances (e.g., co-expression, co-abundance), and (c) conduct rigorous interaction-effect testing. ROMY builds on recent advances in theoretical statistics and double machine learning to ensure robustness and statistical validity. We illustrate the performance of our framework through simulation studies. Availability and implementationromy is an R package available under the GNU GPLv3 license on GitHub: https://github.com/julianhecker/romy.

bioinformatics↗

bioRxiv2025

Context-dependent regulatory variants in Alzheimer's disease

Noncoding genetic variants underlie many complex diseases, yet identifying and interpreting their functional impacts remains challenging. Late-onset Alzheimers disease (LOAD), a polygenic neurodegenerative disorder, exemplifies this challenge. The disease is strongly associated with noncoding variation, including common variants enriched in microglial enhancers and rare variants nominated at loci implicated in neurodevelopment and synaptic function. These variants may perturb regulatory sequences by disrupting transcription factor (TF) motifs or altering local regulatory sequence context, with potential consequences for gene expression and chromatin accessibility. However, assessing their impact is complicated by the context-dependent functions of regulatory sequences, underscoring the need to systematically examine variant effects across diverse tissues, cell types, and cellular states. Here, we combined in vitro and in vivo massively parallel reporter assays (MPRAs) with interpretable machine-learning models to systematically characterize common and rare variants across myeloid and neuron-enriched neural contexts. Parallel profiling of variants in four immune states in vitro and three mouse brain regions in vivo revealed that individual variants can differentially and even oppositely modulate reporter activity across cellular contexts, while a subset showed immune-state-dependent effects. Within the assayed candidate set, the relative effect sizes of common and rare variants reversed between contexts, with common variants showing larger effects in THP-1 macrophages and rare variants in brain tissue. Interpretable sequence-to-function models prioritized context-biased variants and generated motif-level hypotheses, including predicted transcription-factor motif disruption and subtler changes in local motif context. To probe endogenous consequences at a prioritized locus, we used CRISPR interference to perturb a rare-variant-containing enhancer at the SEC63-OSTM1 locus, revealing metabolic and biosynthetic transcriptional programs resembling those induced by SEC63 promoter perturbation, together with a stimulation-specific interferon response. These findings show that LOAD-associated noncoding variants exhibit context-dependent effects on reporter activity and predicted chromatin accessibility, and provide a framework for linking genetic association to regulatory activity, sequence grammar and endogenous transcriptional state.

genomics↗

bioRxiv2025

Molecular hallmarks of excitatory and inhibitory neuronal resilience and resistance to Alzheimer's disease

BackgroundA significant proportion of individuals maintain healthy cognitive function despite having extensive Alzheimers disease (AD) pathology, known as cognitive resilience. Understanding the molecular mechanisms that protect these individuals can identify therapeutic targets for AD dementia. This study aims to define molecular and cellular signatures of cognitive resilience, protection and resistance, by integrating genetics, bulk RNA, and single-nucleus RNA sequencing data across multiple brain regions from AD, resilient, and control individuals. MethodsWe analyzed data from the Religious Order Study and the Rush Memory and Aging Project (ROSMAP), including bulk (n=631) and multi-regional single nucleus (n=48) RNA sequencing. Subjects were categorized into AD, resilient, and control based on {beta}-amyloid and tau pathology, and cognitive status. We identified and prioritized protected cell populations using whole genome sequencing-derived genetic variants, transcriptomic profiling, and cellular composition distribution. ResultsTranscriptomic results, supported by GWAS-derived polygenic risk scores, place cognitive resilience as an intermediate state in the AD continuum. Tissue-level analysis revealed 43 genes enriched in nucleic acid metabolism and signaling that were differentially expressed between AD and resilience. Only GFAP (upregulated) and KLF4 (downregulated) showed differential expression in resilience compared to controls. Cellular resilience involved reorganization of protein folding and degradation pathways, with downregulation of Hsp90 and selective upregulation of Hsp40, Hsp70, and Hsp110 families in excitatory neurons. Excitatory neuronal subpopulations in the entorhinal cortex (ATP8B1+ and MEF2Chigh) exhibited unique resilience signaling through neurotrophin (modulated by LINGO1) and angiopoietin (ANGPT2/TEK) pathways. We identified MEF2C, ATP8B1, and RELN as key markers of resilient excitatory neuronal populations, characterized by selective vulnerability in AD. Protective rare variant enrichment highlighted vulnerable populations, including somatostatin (SST) inhibitory interneurons, validated through immunofluorescence showing co-expression of rare variant associated RBFOX1 and KIF26B in SST+ neurons in the dorsolateral prefrontal cortex. The maintenance of excitatory-inhibitory balance emerges as a key characteristic of resilience. ConclusionsWe identified molecular and cellular hallmarks of cognitive resilience, an intermediate state in the AD continuum. Resilience mechanisms include preservation of neuronal function, maintenance of excitatory/inhibitory balance, and activation of protective signaling pathways. Specific excitatory neuronal populations appear to play a central role in mediating cognitive resilience, while a subset of vulnerable SST interneurons likely provide compensation against AD-associated dysregulation. This study offers a framework to leverage natural protective mechanisms to mitigate neurodegeneration and preserve cognition in AD.

neuroscience↗

bioRxiv2024

The developing Human Connectome Project fetal functional MRI release: Methods and data structures

Recent advances in fetal fMRI present a new opportunity for neuroscience to study functional human brain connectivity at the time of its emergence. Progress in the field however has been hampered by the lack of openly available datasets that can be exploited by researchers across disciplines to develop methods that would address the unique challenges associated with imaging and analysing functional brain in utero, such as unconstrained head motion, dynamically evolving geometric distortions, or inherently low signal-to-noise ratio. Here we describe the developing Human Connectome Projects release of the largest open access fetal fMRI dataset to date, containing 275 scans from 255 fetuses and spanning the period of 20.86 to 38.29 post-menstrual weeks. We present a systematic approach to its pre-processing, implementing multi-band soft SENSE reconstruction, dynamic distortion corrections via phase unwrapping method, slice-to-volume reconstruction and a tailored temporal filtering model, with attention to the prominent sources of structured noise in the in utero fMRI. The dataset is accompanied with an advanced registration infrastructure, enabling group-level data fusion, and contains outputs from the main intermediate processing steps. This allows for various levels of data exploration by the imaging and neuroscientific community, starting from the development of robust pipelines for anatomical and temporal corrections to methods for elucidating the development of functional connectivity in utero. By providing a high-quality template for further method development and benchmarking, the release of the dataset will help to advance fetal fMRI to its deserved and timely place at the forefront of the efforts to build a life-long connectome of the human brain.

neuroscience↗

bioRxiv2024

Polygenic hazard score models for the prediction of Alzheimer's free survival using the lasso for Cox's proportional hazards model

The prediction of the susceptibility of an individual to a certain disease is an important and timely research area. An established technique is to estimate the risk of an individual with the help of an integrated risk model, that is a polygenic risk score with added epidemiological covariates. However, integrated risk models do not capture any time dependence, and may provide a point estimate of the relative risk with respect to a reference population. The aim of this work is twofold. First, we explore and advocate the idea of predicting the time dependent hazard and survival (defined as disease free time) of an individual for the onset of a disease. This provides a practitioner with a much more differentiated view of the absolute survival as a function of time. Second, to compute the time dependent risk of an individual, we use published methodology to fit a Coxs proportional hazard model to data from a genetic SNP study of time to Alzheimers disease (AD) onset, using the lasso to incorporate further epidemiological variables such as sex, APOE (apolipoprotein E, a genetic risk factor for AD) status, ten leading principal components, and selected genomic loci. We apply the lasso for Coxs proportional hazards to a dataset of 6792 AD patients (composed of 4102 cases and 2690 controls) and 87 covariates. We demonstrate that fitting a lasso model for Coxs proportional hazards allows one to obtain more accurate survival curves than with state-of-the-art (likelihood-based) methods. Moreover, the methodology allows one to obtain personalized survival curves for a patient, thus giving a much more differentiated view of the expected progression of a disease than the view offered by integrated risk models. The runtime to compute personalized survival curves is under a minute for the entire dataset of AD patients, thus enabling it to handle datasets with 60, 000 to 100, 000 subjects in less than one hour.

bioinformatics↗

bioRxiv2022

Fast computation of principal components of genomic similarity matrices

The computation of a similarity measure for genomic data, for instance using the (genomic) covariance matrix, the Jaccard matrix, or the genomic relationship matrix (GRM), is a standard tool in computational genetics. The principal components of such matrices are routinely used to correct for biases in, for instance, linear regressions. However, the calculation of both a similarity matrix and its singular value decomposition (SVD) are computationally intensive. The contribution of this article is threefold. First, we demonstrate that the calculation of three matrices (the genomic covariance matrix, the weighted Jaccard matrix, and the genomic relationship matrix) can be reformulated in a unified way which allows for an exact, faster SVD computation. An exception is the Jaccard matrix, which does not have a structure applicable for the fast SVD computation. An exact algorithm is proposed to compute the principal components of the genomic covariance, weighted Jaccard, and genomic relationship matrices. The algorithm is adapted from an existing randomized SVD algorithm and ensures that all computations are carried out in sparse matrix algebra. Second, an approximate Jaccard matrix is introduced to which the fast SVD computation is applicable. Third, we establish guaranteed theoretical bounds on the distance (in L2 norm and angle) between the principal components of the Jaccard matrix and the ones of our proposed approximation, thus putting the proposed Jaccard approximation on a solid mathematical foundation. We illustrate all computations on both simulated data and data of the 1000 Genome Project, showing that the approximation error is very low in practice.

bioinformatics↗

bioRxiv2022

Unsupervised outlier detection applied to SARS-CoV-2 nucleotide sequences can identify sequences of common variants and other variants of interest

As of June 2022, the GISAID database contains more than one million SARS-CoV-2 genomes, including several thousand nucleotide sequences for the most common variants such as delta or omicron. These SARS-CoV-2 strains have been collected from patients around the world since the beginning of the pandemic. We start by assessing the similarity of all pairs of nucleotide sequences using the Jaccard index and principal component analysis. As shown previously in the literature, an unsupervised cluster analysis applied to the SARS-CoV-2 genomes results in clusters of sequences according to certain characteristics such as their strain or their clade. Importantly, we observe that nucleotide sequences of common variants are often outliers in clusters of sequences stemming from variants identified earlier on during the pandemic. Motivated by this finding, we are interested in applying outlier detection to nucleotide sequences. We demonstrate that nucleotide sequences of common variants (such as alpha, delta, or omicron) can be identified solely based on a statistical outlier criterion. We argue that outlier detection might be a useful surveillance tool to identify emerging variants in real time as the pandemic progresses.

bioinformatics↗

bioRxiv2021

Unsupervised genome-wide cluster analysis: nucleotide sequences of the omicron variant of SARS-CoV-2 are similar to sequences from early 2020

The GISAID database contains more than 1,000,000 SARS-CoV-2 genomes, including sequences of the recently discovered SARS-CoV-2 omicron variant and of prior SARS-CoV-2 strains that have been collected from patients around the world since the beginning of the pandemic. We applied unsupervised cluster analysis to the SARS-CoV-2 genomes, assessing their similarity at a genome-wide level based on the Jaccard index and principal component analysis. Our analysis results show that the omicron variant sequences are most similar to sequences that have been submitted early in the pandemic around January 2020. Furthermore, the omicron variants in GISAID are spread across the entire range of the first principal component, suggesting that the strain has been in circulation for some time. This observation supports a long-term infection hypothesis as the omicron strain origin.

bioinformatics↗

bioRxiv2021

A robust and adaptive framework for interaction testing in quantitative traits between multiple genetic loci and exposure variables

The identification and understanding of gene-environment interactions can provide insights into the pathways and mechanisms underlying complex diseases. However, testing for gene-environment interaction remains a challenge since statistical power is often limited, the specification of environmental effects is nontrivial, and such misspecifications can lead to false positive findings. To address the lack of statistical power, recent methods aim to identify interactions on an aggregated level using, for example, polygenic risk scores. While this strategy increases power to detect interactions, identifying contributing key genes and pathways is difficult based on these global results. Here, we propose RITSS (Robust Interaction Testing using Sample Splitting), a gene-environment interaction testing framework for quantitative traits that is based on sample splitting and robust test statistics. RITSS can incorporate multiple genetic variants and/or multiple environmental factors. Using sample splitting, a screening step enables the selection and combination of potential interactions into scores with improved interpretability, based on the users unrestricted choices for statistical/machine learning approaches. In the testing step, the application of robust test statistics minimizes the susceptibility of the results to main effect misspecifications. Using extensive simulation studies, we demonstrate that RITSS controls the type 1 error rate in a wide range of scenarios. In an application to lung function phenotypes and human height in the UK Biobank, RITSS identified genome-wide significant interactions with subcomponents of genetic risk scores. While the contributing single variant interactions are moderate, our analysis results indicate interesting interaction patterns that result in strong aggregated signals that provide further insights into gene-environment interaction mechanisms.

bioinformatics↗

bioRxiv2021

A smoothed version of the Lassosum penalty for fitting integrated risk models

Polygenic risk scores are a popular means to predict the disease risk or disease susceptibility of an individual based on its genotype information. When adding other important epidemiological covariates such as age or sex, we speak of an integrated risk model. Methodological advances for fitting more accurate integrated risk models are of immediate importance to improve the precision of risk prediction, thereby potentially identifying patients at high risk early on when they are still able to benefit from preventive steps/interventions targeted at increasing their odds of survival, or at reducing their chance of getting a disease in the first place. This article proposes a smoothed version of the "Lassosum" penalty used to fit polygenic risk scores and integrated risk models. The smoothing allows one to obtain explicit gradients everywhere for efficient minimization of the Lassosum objective function while guaranteeing bounds on the accuracy of the fit. An experimental section on both Alzheimers disease and COPD (chronic obstructive pulmonary disease) demonstrates the increased accuracy of the proposed smoothed Lassosum penalty compared to the original Lassosum algorithm, allowing it to draw equal with state-of-the-art methodology such as LDpred2 when evaluated via the AUC (area under the ROC curve) metric.

bioinformatics↗

bioRxiv2020

TMEM106B and CPOX are genetic determinants of cerebrospinal fluid Alzheimer's disease biomarker levels

BackgroundNeurofilament light (NF-L), chitinase-3-like protein 1 (YKL-40), and neurogranin (Ng) are utilized as biomarkers for Alzheimers disease (AD), to monitor axonal damage, astroglial activation, and synaptic degeneration, respectively. Here we performed genome-wide association study (GWAS) analyses using all three biomarkers as outcome. MethodsDNA and cerebrospinal fluid (CSF) samples originated from the European Medical Information Framework AD Multimodal Biomarker Discovery (EMIF-AD MBD) study. Overlapping genotype/phenotype data were available for n=671 (NF-L), 677 (YKL-40), and 672 (Ng) individuals. GWAS analyses applied linear regression models adjusting for relevant covariates. FindingsWe identify novel genome-wide significant associations with markers in TMEM106B and CSF levels of NF-L. Additional novel signals were observed with DNA variants in CPOX and CSF levels of YKL-40. Lastly, we confirmed previous work suggesting that YKL-40 levels are regulated by cis protein quantitative trait loci (pQTL) in CHI3L1. InterpretationOur study provides important new insights into the genetic architecture underlying inter-individual variation in all three tested AD-related CSF biomarkers. In particular, our data shed light on the sequence of events regarding the initiation and progression of neuropathological processes relevant in AD.

genomics↗

bioRxiv2020

locStra: Fast analysis of regional/global stratification in whole genome sequencing (WGS) studies

locStra is an R-package for the analysis of regional and global population stratification in whole genome sequencing studies, where regional stratification refers to the substructure defined by the loci in a particular region on the genome. Population substructure can be assessed based on the genetic covariance matrix, the genomic relationship matrix, and the unweighted/weighted genetic Jaccard similarity matrix. Using a sliding window approach, the regional similarity matrices are compared to the global ones, based on user-defined window sizes and metrics, e.g. the correlation between regional and global eigenvectors. An algorithm for the specification of the window size is provided. As the implementation fully exploits sparse matrix algebra and is written in C++, the analysis is highly efficient. Even on single cores, for realistic study sizes (several thousand subjects, several million RVs per subject), the runtime for the genome-wide computation of all regional similarity matrices does typically not exceed one hour, enabling an unprecedented investigation of regional stratification across the entire genome. The package is applied to three WGS studies, illustrating the varying patterns of regional substructure across the genome and its beneficial effects on association testing.

genetics↗

bioRxiv2019

Estimating the effective sample size in association studies of quantitative traits

The effective sample size (ESS) is a quantity estimated in genome-wide association studies (GWAS) with related individuals and/or linear mixed models used in analysis. ESS originally measured relative power in family-based GWAS and has recently become important for correcting GWAS summary statistics in post-GWAS analyses. However, existing ESS approaches have been overlooked and based on empirical estimation. This work presents an analytical form of ESS in mixed-model GWAS of quantitative traits, which is derived using the expectation of quadratic form and validated in extensive simulations. We illustrate the performance and relevance of our ESS estimator in common GWAS scenarios and analytically show that (i) family-based studies are consistently underpowered compared to studies of unrelated individuals of the same sample size; (ii) conditioning on polygenic genetic effect by linear mixed models boosts power; and (iii) power of detecting gene-environment interaction can be substantially gained or lost in family-based designs depending on exposure distribution. We further analyze UK Biobank dataset in two samples of 336,347 unrelated and 68,910 related individuals. Analysis in unrelated individuals reveals a high accuracy of our ESS estimator compared to the existing empirical approach; and analysis of related individuals suggests that the loss in effective sample size due to relatedness is at most 0.94x. Overall, we provide an analytical form of ESS for guiding GWAS designs and processing summary statistics in post-GWAS analyses.

genetics↗

bioRxiv2019

Genome-wide association study of Alzheimer’s disease CSF biomarkers in the EMIF-AD Multimodal Biomarker Discovery dataset

Alzheimers disease (AD) is the most prevalent neurodegenerative disorder and the most common form of dementia in the elderly. Susceptibility to AD is considerably determined by genetic factors which hitherto were primarily identified using case-control designs. Elucidating the genetic architecture of additional AD-related phenotypic traits, ideally those linked to the underlying disease process, holds great promise in gaining deeper insights into the genetic basis of AD and in developing better clinical prediction models. To this end, we generated genome-wide single-nucleotide polymorphism (SNP) genotyping data in 931 participants of the European Medical Information Framework Alzheimers Disease Multimodal Biomarker Discovery (EMIF-AD MBD) sample to search for novel genetic determinants of AD biomarker variability. Specifically, we performed genome-wide association study (GWAS) analyses on 16 traits, including 14 measures of amyloid-beta (A{beta}) and tau-protein species in the cerebrospinal fluid (CSF). In addition to confirming the well-established effects of apolipoprotein E (APOE) on diagnostic outcome and phenotypes related to A{beta}42, we detected novel potential signals in the zinc finger homeobox 3 (ZFHX3) for CSF-A{beta}38 and CSF-A{beta}40 levels, and confirmed the previously described sex-specific association between SNPs in geminin coiled-coil domain containing (GMNC) and CSF-tau. Utilizing the results from independent case-control AD GWAS to construct polygenic risk scores (PRS) revealed that AD risk variants only explain a small fraction of CSF biomarker variability. In conclusion, our study represents a detailed first account of GWAS analyses on CSF-A{beta} and -tau related traits in the EMIF-AD MBD dataset. In subsequent work, we will utilize the genomics data generated here in GWAS of other AD-relevant clinical outcomes ascertained in this unique dataset.

genomics↗

bioRxiv2018

Expanded genetic landscape of chronic obstructive pulmonary disease reveals heterogeneous cell type and phenotype associations

Chronic obstructive pulmonary disease (COPD) is the leading cause of respiratory mortality worldwide. Genetic risk loci provide novel insights into disease pathogenesis. To broaden COPD genetic risk loci discovery and identify cell type and phenotype associations we performed a genome-wide association study in 35,735 cases and 222,076 controls from the UK Biobank and additional studies from the International COPD Genetics Consortium. We identified 82 loci with P value < 5x10-8; 47 were previously described in association with either COPD or population-based lung function. Of the remaining 35 novel loci, 13 were associated with lung function in 79,055 individuals from the SpiroMeta consortium. Using gene expression and regulation data, we identified enrichment for loci in lung tissue, smooth muscle and alveolar type II cells. We found 9 shared genomic regions between COPD and asthma and 5 between COPD and pulmonary fibrosis. COPD genetic risk loci clustered into groups of quantitative imaging features and comorbidity associations. Our analyses provide further support to the genetic susceptibility and heterogeneity of COPD.

genetics↗

bioRxiv2017

Reporting correct p-values in VEGAS analyses

VEGAS (versatile gene-based association study) is a popular methodological framework to perform gene-based tests based on summary statistics from single-variant analyses. The approach incorporates linkage disequilibrium information from reference panels to account for the correlation of test statistics. The gene-based test can utilize three different types of tests. In 2015, the improved framework VEGAS2, using more detailed reference panels, was published. Both versions provide user-friendly web- and offline-based tools for the analysis. However, the implementation of the popular top-percentage test is erroneous in both versions. The p-values provided by VEGAS2 are deflated/anti-conservative. Based on real data examples, we demonstrate that this can increase substantially the rate of false positive findings and can lead to inconsistencies between different test options. We also provide code that allows the user of VEGAS to compute correct p-values.

genetics↗