Biology subjects

Smirnov, D.

Publications and source records attributed to Smirnov, D..

11 recordsLinked to original sources

bioRxiv2026

A systematic evaluation of SIRT6 as transcriptomic biomarker of aging

SIRT6 is a NAD+-dependent sirtuin that plays central roles in chromatin regulation, DNA repair, telomere maintenance, metabolic homeostasis, and inflammatory control. Although SIRT6 have long been implicated in aging because of its association with several hallmarks of aging, the available evidence remains largely context-dependent and mechanistic, limiting the interpretation of SIRT6 as a robust and evolutionarily conserved biomarker of aging. To comprehensively investigate the role of SIRT6 as an aging biomarker, we established SIRT6.db, a multi-species transcriptomic resource that integrates SIRT6-targeted perturbation experiments across diverse biological systems and organisms form all available SIRT6-related publications collected via a large-scale textual analysis of the SIRT6 literature using topic modeling. Based on this database, we identified both species-specific and evolutionarily conserved transcriptional and functional signatures associated with SIRT6 perturbation and established their relevance to hallmarks of aging. We showed that SIRT6 expression is generally stable during normal aging, but becomes dysregulated in Alzheimer's disease in cell type and stage-specific manner, highlighting the context dependence of its potential as a transcriptomic biomarker of aging.

bioinformatics↗

bioRxiv2026

Benchmarking single-cell foundation models for real-world RNA-seq data integration

Single-cell foundation models enable reusable representations and streamlined analysis workflows, yet rigorous evaluation of their performance and robustness in real-world pharmaceutical settings remain underexplored. Here, we benchmarked leading single-cell foundation models (scGPT; scGPT_CP, a continually pretrained checkpoint of scGPT; scFoundation; scMulan; CellFM) against established baseline methods (scVI; Harmony) for data integration using over 1.5 million cells from clinical and preclinical samples. Performance was assessed using well-established and complementary metrics for technical correction and biological structure preservation. We further introduced robustness-oriented rankings to summarize metric trade-offs and quantify performance consistency across datasets and evaluation settings. Our findings show that fine-tuning improved technical correction performance; among the foundation models, fine-tuned scGPT_CP performed best. However, the baseline scVI was the top overall performer, ranking first by our multi-metric Leximax ranking and achieving the highest Pareto Front-1 hit. Collectively, our study provides practical insights for adapting foundation models to real-world drug design and development.

bioinformatics↗

bioRxiv2026

Aging in Fast-Forward: An Inducible SIRT6 Deficiency model as a Lens on Brain Aging and Neurodegeneration

Aging and neurodegeneration occur gradually, making in vitro modeling challenging and costly. We generated a time-resolved, reversible neuron-like aging model by gradually depleting SIRT6. Within three weeks, their transcriptomes recapitulated brain-aging signatures and hallmarks. RNA-seq revealed clusters of nonlinear changes and predicted oscillatory DNA-damage and apoptotic programs, allowing stress response and adaptation. SIRT6 loss led to nuclear envelope breakdown and micronuclei accumulation, resembling accelerated-aging phenotypes. Some defects could be rescued by re-expressing SIRT6. As a tool for discovery, we uncovered disrupted nucleocytoplasmic transport as an aging pathway shared with neurodegenerative disorders. When SIRT6 depletion reached 30 days, transcriptional changes correlated with those in Alzheimers patients, but reversed after SIRT6 re-expression. Pathways associated with shSIRT6 and healthy aging became anticorrelated with AD, pointing to critical signatures. Our affordable and easy-to-use model captures key molecular features of aging, distinguishes physiological from disease-linked changes, and accelerates mechanistic discovery in a controllable, neuron-based system. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=140 SRC="FIGDIR/small/705977v1_ufig1.gif" ALT="Figure 1"> View larger version (38K): org.highwire.dtl.DTLVardef@181c9d1org.highwire.dtl.DTLVardef@1e8be3dorg.highwire.dtl.DTLVardef@121d345org.highwire.dtl.DTLVardef@1b5589e_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOGraphical Abstract:C_FLOATNO A Novel Reversible Cellular Model of Gradual SIRT6 Decline Reveals Transcriptomic Transitions from Aging to Neurodegeneration. (1) We developed a neuron-like cellular system with gradual, inducible SIRT6 depletion to mimic the aging process. The model enables: (2) Recapitulation of hallmark features of aging, including genomic instability, cellular stress, and nuclear integrity loss. (3) Identification of novel SIRT6-regulated pathways, including impaired nucleocytoplasmic transport. (4) Clustering of dynamic, non-linear gene expression trajectories. (5) Distinction between reversible and irreversible changes following SIRT6 re-expression. (6) Transcriptomic comparison with human brain aging and AD revealing divergence points between healthy and pathological aging. C_FIG

molecular biology↗

bioRxiv2026

Benchmarking Next-Generation Sequencing Platforms: A Comprehensive Comparison Of Single-Cell RNA-Seq from Ultima UG 100 vs. Illumina NovaSeq X Plus

Single-cell RNA sequencing (scRNA-seq) has become an essential technology for dissecting cellular heterogeneity in complex biological systems, with Illumina platforms serving as the dominant NGS provider1-3. The recent emergence of the Ultima Genomics UG 100 platform offers a high-throughput and potentially more cost-effective alternative, warranting a rigorous comparative analysis. Here, we conducted a comprehensive head-to-head comparison of the Ultima Genomics UG 100 and Illumina NovaSeq X Plus platforms, generating deep-coverage scRNA-seq data from human PBMCs and stimulated T-cell samples. Our multi-level analysis revealed highly concordant performance in both cell type identification and gene expression. We found that both platforms robustly captured all major immune lineages and demonstrated good transcriptional agreement. Minor observed discrepancies were primarily correlated with low UMI counts, suggesting stochastic biological effects rather than technical variance. Overall, our study establishes that the Ultima UG 100 delivers performance highly comparable to the Illumina NovaSeq X Plus for single-cell transcriptomics.

bioinformatics↗

bioRxiv2025

The Divergent 3D Genome Landscapes of Aging and Neurodegenerative mouse models

Chromatin structure is essential for gene regulation and genome stability, and neurons must preserve their 3D genome organization throughout life. This structure gradually deteriorates with aging and is further disrupted in neurodegenerative diseases. To investigate whether aging and neurodegeneration share early chromatin changes or diverge, we performed Hi-C on cortical neurons from adult, aged, and brain-specific SIRT6-knockout (S6-KO) mice, and compared them to the CK-p25 Alzheimers disease model results stratified by {gamma}H2AX levels. All models showed early features such as weakened interactions between chromosomes in expanded nuclei, A-to-B compartment shifts, and loss of architectural loops. However, aged neurons retained TAD prominence, short-range interactions, and enhancer-promoter loops, while pathological models showed reduced TAD prominence, shorter loops, and disorganized A-B mixing. These changes correlate with increased DNA damage. Our findings suggest that aging represents a "primed" state, where chromatin regulation begins to erode, but further stressors like DNA damage are associated with progression toward neurodegenerative breakdown.

molecular biology↗

bioRxiv2025

PROTRIDER: Protein abundance outlier detection from mass spectrometry-based proteomics data with a conditional autoencoder

Structured abstractO_ST_ABSMotivationC_ST_ABSDetection of gene regulatory aberrations enhances our ability to interpret the impact of inherited and acquired genetic variation for rare disease diagnostics and tumor characterization. While numerous methods for calling RNA expression outliers from RNA-sequencing data have been proposed, the establishment of protein expression outliers from mass spectrometry data is lacking. ResultsHere, we propose and assess various modeling approaches to call protein expression outliers across three datasets from rare disease diagnostics and oncology. We use as independent evidence the enrichment for outlier calls in matched RNA-seq samples and the enrichment for rare variants likely disrupting protein expression. We show that controlling for hidden confounders and technical covariates, while simultaneously modeling the occurrence of missing values, is largely beneficial and can be achieved using conditional autoencoders. Moreover, we find that the differences between experimental and fitted log-transformed intensities by such models exhibit heavy tails that are poorly captured with the Gaussian distribution and report stronger statistical calibration when instead using the Students t-distribution. Our resulting method, PROTRIDER, outperformed baseline approaches based on raw log-intensities Z-scores or on differential expression analysis with limma. The application of PROTRIDER reveals significant enrichments of AlphaMissense pathogenic variants in protein expression outliers. Overall, PROTRIDER provides a method to confidently identify aberrantly expressed proteins applicable to rare disease diagnostics and cancer proteomics. Availability and ImplementationPROTRIDER is freely available at github.com/gagneurlab/PROTRIDER and also available on Zenodo under the DOI zenodo.15569781. ContactJulien Gagneur: gagneur at in.tum.de

bioinformatics↗

bioRxiv2023

optimalTAD: annotation of topologically associating domains based on chromatin marks enrichment

In many eukaryotes, chromosomes are organized as strings of spatially segregated Topologically Associating Domains (TADs), characterized by a substantially increased frequency of interactions within them. Boundaries of TADs are highly enriched in histone acetylation chromatin marks and occupied binding sites of architectural proteins, highlighting the functional role of TADs in the regulation of gene expression. While many computational approaches have been developed for TAD identification, it remains challenging because of their nested structure, resulting in weakly overlapping sets of TADs at different scales. Here, we propose a novel algorithm optimalTAD for identifying the optimal set of TADs based on epigenetic marks enrichment. Assuming that the most dramatic enrichment corresponds to the best annotation of TAD boundaries, our algorithm optimizes TAD calling parameters by maximizing the difference in chromatin mark levels between TADs and their boundaries. Using this algorithm, we annotated TADs in multiple publicly available fruit fly and mammalian Hi-C datasets and identified a set of epigenetic marks that are best suited for TAD prediction. Through the analysis of diverse organisms and cell types with distinct underlying principles of TAD organization, we have shown that optimalTAD is a universal tool suitable for studying TAD structure, functions, and properties unique to specific cell types and organisms. optimalTAD is freely available at GitHub: https://github.com/cosmoskaluga/optimalTAD. Key PointsO_LIWe assume that the most dramatic enrichment of epigenetic marks corresponds to the best annotation of TAD boundaries. C_LIO_LIOur algorithm optimizes TAD calling parameters by maximizing the difference in chromatin mark levels between TADs and their boundaries. C_LIO_LIoptimalTAD is a universal tool that is applicable for studying TAD characteristics in diverse organisms and cell types. C_LIO_LIoptimalTAD enables the identification of a specific set of epigenetic marks that are most suitable for annotating TADs. C_LI

bioinformatics↗

bioRxiv2022

SIRT6 is a key regulator of mitochondrial function in the brain

SIRT6 is implicated in DNA repair, telomere maintenance, glucose and lipid metabolism and, importantly, it has critical roles in the brain ranging from its development to neurodegeneration. In this work, we combined transcriptomics and metabolomics approaches to characterize the functions of SIRT6 in mice brains. Our analysis revealed that SIRT6 is a critical regulator of mitochondrial activity in the brain. In its absence, there is a mitochondrial deficiency with a global downregulation of mitochondria-related genes and pronounced changes in metabolites content. We predict that SIRT6 can affect mitochondrial functions through its interaction with the transcription factor YY1 that, together, regulate mitochondrial gene expression. Moreover, SIRT6 target genes include SIRT3 and SIRT4, which are significantly downregulated in SIRT6-deficient brains. Our results demonstrate that the lack of SIRT6 leads to decreased mitochondrial gene expression and metabolomic changes of TCA cycle byproducts, including increased ROS production, reduced mitochondrial number, and impaired membrane potential that can be partially rescued by restoring SIRT3 and 4 levels. Importantly, the changes observed in SIRT6 deficient brains are observed in brains of aging people, but the overlapping is greater in patients with Alzheimers, Parkinsons, Huntingtons, and Amyotrophic lateral sclerosis disease. Overall, our results suggest that reduced levels of SIRT6 in the aging brain and neurodegeneration could initiate mitochondrial dysfunction by altering gene expression, ROS production and mitochondrial decay.

cell biology↗

bioRxiv2020

Solving musculoskeletal biomechanics with machine learning

Deep learning is a relatively new computational technique for the description of the musculoskeletal dynamics. The experimental relationships of muscle geometry in different postures are the high-dimensional spatial transformations that can be approximated by relatively simple functions, which opens the opportunity for machine learning applications. In this study, we challenged general machine learning algorithms with the problem of approximating the posture-dependent moment arm and muscle length relationships of the human arm and hand muscles. We used two types of algorithms, light gradient boosting machine (LGB) and fully connected artificial neural network (ANN) solving the wrapping kinematics of 33 muscles spanning up to six degrees of freedom (DOF) each for the arm and hand model with 18 DOFs. The input-output training and testing datasets were generated by our previous phenomenological model based on the autogenerated polynomial structures (Sobinov et al., 2019). Both models achieved a similar level of errors: ANN model errors were 0.08{+/-}0.05% for muscle lengths and 0.53{+/-}0.29% for moment arms, and LGB model made similar errors--0.18{+/-}0.06% and 0.13{+/-}0.07%, respectively. LGB model reached the training goal with only 10^3 samples, while ANN required 10^6 samples; however, LGB models were about 39 slower than ANN models in the evaluation. The sufficient performance of developed models demonstrates the future applicability of machine learning for musculoskeletal transformations in a variety of applications, such as in advanced powered prosthetics. Author SummaryThe accurate decoding of arm and hand motor intent from biological signals remains a key challenge. Solving this task with machine learning requires vast posture- and task-dependent data for identifying structural and functional parameters within dynamic musculoskeletal relationships. This problem is related to the curse of dimensionality where the processing complexity grows exponentially with the number of degrees of freedom described by the model. Here, we developed a tool based on artificial neural networks (ANN) to solve the kinematic transformation from posture to muscle path length and muscle moment arms. We used an accurate model of posture-dependent muscle moment arms and length to train and test the ability of ANN to solve this high-dimensional and computationally intense transformation and compare it to the boosted decision tree approach. We demonstrated that model-driven training is an efficient method to handle the encoding of high-dimensional musculoskeletal relationships. Adding muscles to the transformation, which increases the input-output complexity, does not reduce the prediction accuracy and does not require the increase in the number of elements within the network demonstrating the viability of this approach for applications using musculoskeletal biomechanics.

neuroscience↗

bioRxiv2020

Classification of emotions based on functional connectivity patterns of the human brain

Neurophysiological and psychological models posit that emotions depend on connections across wide-spread corticolimbic circuits. While previous studies using pattern recognition on neuroimaging data have shown differences between various discrete emotions in brain activity patterns, less is known about the differences in functional connectivity. Thus, we employed multivariate pattern analysis on functional magnetic resonance imaging data (i) to develop a pipeline for applying pattern recognition in functional connectivity data, and (ii) to test whether connectivity signatures differ across emotions. Six emotions (anger, fear, disgust, happiness, sadness, and surprise) and a neutral state were induced in 16 participants using one-minute-long emotional narratives with natural prosody while brain activity was measured with functional magnetic resonance imaging (fMRI). We computed emotion-wise connectivity matrices both for whole-brain connections and for 10 previously defined functionally connected brain subnetworks, and trained an across-participant classifier to categorize the emotional states based on whole-brain data and for each subnetwork separately. The whole-brain classifier performed above chance level with all emotions except sadness, suggesting that different emotions are characterized by differences in large-scale connectivity patterns. When focusing on the connectivity within the 10 subnetworks, classification was successful within the default mode system and for all emotions. We conclude that functional connectivity patterns consistently differ across different emotions particularly within the default mode system.

neuroscience↗

bioRxiv2019

Statistical Pattern Recognition Reveals Shared Neural Signatures for Displaying and Recognizing Specific Facial Expressions

Human neuroimaging and behavioural studies suggest that somatomotor "mirroring" of seen facial expressions may support their recognition. Here we show that viewing specific facial expressions triggers the representation corresponding to that expression in the observers brain. Twelve healthy female volunteers underwent two separate fMRI sessions: one where they observed and another where they displayed three types of basic facial expressions (joy, anger and disgust). Pattern classifier based on Bayesian logistic regression was trained to classify facial expressions i) within modality (trained and tested with data recorded while observing or displaying expressions) and ii) between modalities (trained with data recorded while displaying expressions and tested with data recorded while observing the expressions). Cross-modal classification was performed in two ways: with and without functional realignment of the data across observing / displaying conditions. All expressions could be accurately classified within and also across modalities. Brain regions contributing most to cross-modal classification accuracy included primary motor and somatosensory cortices. Functional realignment led to only minor increases in cross-modal classification accuracy for most of the examined ROIs. Substantial improvement was observed in the occipito-ventral components of the core system for facial expression recognition. Altogether these results support the embodied emotion recognition model and show that expression-specific somatomotor neural signatures could support facial expression recognition.

neuroscience↗