Biology subjects

Wang, M. D.

Publications and source records attributed to Wang, M. D..

At least 19 recordsLinked to original sources

bioRxiv2025

Data-Driven Symbolic Higher-Order Epistasis Discovery with Kolmogorov-Arnold Networks

Many human diseases are polygenic conditions that arise from a complex interplay of interactions between multiple genes at different loci, but currently most Genome-Wide Association Studies (GWAS) largely only consider the main additive effects of single nucleotide polymorphisms (SNPs), resulting in a missing heritability problem in some complex traits. Identifying non-additive interactions, or epistasis, at a higher-order could aid in filling this gap, but it is computationally difficult due to the massive search space involved. Current epistasis detection approaches struggle with noncartesian higher order interactions and lack inherent explainability. We present a novel deep learning (DL) approach, EPIstasis Discovery with Kolmogorov-Arnold Networks (EPIK), a data-driven, modular, and symbolically representable framework. We also introduce a novel approach for higher-order XOR (a non-Cartesian type) interaction detection, utilized in EPIKs XOR detection module. EPIK slightly outperforms other DL approaches on simulated pure epistasis interactions benchmark in average F1 score. It outperforms other, general, traditional epistasis detection approaches on simulated mixed epistasis detection datasets and real-world GWAS datasets of Arabidopsis Thaliana. Finally, EPIK recovers a known gene interaction between MAPT and WNT3 for Parkinsons Disease (PD) while also suggesting a more complex interaction between MAPT, WNT3, and another gene, KANSL1.

genomics↗

bioRxiv2025

The Eukaryotic homology search complex distorts donor DNA structure to probe for homology

Homologous recombination (HR) is a DNA double-strand break repair pathway that facilitates genetic exchange and protects damaged replication forks during DNA synthesis. As a template-based repair process, the successful repair of a double-strand break depends on locating suitable homology from a donor DNA sequence elsewhere in the genome. In eukaryotes, Rad51 catalyzes the homology search in coordination with the ATP-dependent motor protein Rad54. The mechanism by which these two proteins regulate forces on dsDNA substrates during homology search remains unknown. Here, we have utilized single-molecule magnetic tweezers and optical trapping methods to monitor remodeling of the DNA template during the homology search. We find that the activity of Rad51 and Rad54 remodels the donor DNA substrate to control the association and dissociation of Rad51-ssDNA filaments in the absence of DNA homology. This mechanism occurs through the application of both linear (tension) and rotational (torsion) forces on the donor DNA. Finally, failure of Rad54 to act processively disrupts target selection in vivo. This study provides a basic understanding of how motorized homology search manipulates the donor DNA during the search for a suitable repair template. Significance StatementHomologous recombination (HR) is a double-strand DNA break repair pathway that utilizes a template-based target search process to locate a suitable homologous DNA sequence in the genome, thereby initiating DNA repair. Called the homology search, in eukaryotes, this process is carried out by the RecA family member Rad51. During the homology search, Rad51 collaborates with the motor protein Rad54 to identify and interrogate homologous DNA sequences within the genome. In this study, we have measured the forces applied by the combination of Rad51 and Rad54 to the donor DNA duplex. These measurements reveal a coordinated effort by these motor proteins to remodel donor DNA to probe for homology, shedding new light on how template-based homology searches interrogate the DNA strands.

biochemistry↗

bioRxiv2024

Chromatin Buffers Torsional Stress During Transcription

Transcription through chromatin under torsion represents a fundamental problem in biology. Pol II must overcome nucleosome obstacles and, because of the DNA helical structure, must also rotate relative to the DNA, generating torsional stress. However, there is a limited understanding of how Pol II transcribes through nucleosomes while supercoiling DNA. In this work, we developed methods to visualize Pol II rotation of DNA during transcription and determine how torsion slows down the transcription rate. We found that Pol II stalls at {+/-} 9 pN{middle dot}nm torque, nearly sufficient to melt DNA. The stalling is due to extensive backtracking, and the presence of TFIIS increases the stall torque to + 13 pN{middle dot}nm, making Pol II a powerful rotary motor. This increased torsional capacity greatly enhances Pol IIs ability to transcribe through a nucleosome. Intriguingly, when Pol II encounters a nucleosome, nucleosome passage becomes more efficient on a chromatin substrate than on a single-nucleosome substrate, demonstrating that chromatin efficiently buffers torsional stress via its torsional mechanical properties. Furthermore, topoisomerase II relaxation of torsional stress significantly enhances transcription, allowing Pol II to elongate through multiple nucleosomes. Our results demonstrate that chromatin greatly reduces torsional stress on transcription, revealing a novel role of chromatin beyond the more conventional view of it being just a roadblock to transcription.

biophysics↗

bioRxiv2024

Human Topoisomerase IIα Promotes Chromatin Condensation Via a Phase Transition

Topoisomerase II (topo II) enzymes are essential enzymes known to resolve topological entanglements during DNA processing. Curiously, while yeast expresses a single topo II, humans express two topo II isozymes, topo II and topo II{beta}, which share a similar catalytic domain but differ in their intrinsically disordered C-terminal domains (CTDs). During mitosis, topo II and condensin I constitute the most abundant chromosome scaffolding proteins essential for chromosome condensation. However, how topo II enables this function is poorly understood. Here, we discovered a new and functionally distinct role for human topo II - it condenses DNA and chromatin at a low topo II concentration (100 pM or less) during a polymer-collapse phase transition. The removal of the topo II CTDs effectively abolishes its condensation ability, indicating that the condensation is mediated by the CTDs. Although topo II{beta} can also perform condensation, it is about 4-fold less effective. During the condensation, topo II-DNA condensates form along DNA, working against a DNA tension of up to 1.5 pN, greater than that previously reported for yeast condensin. In addition, this condensation does not require ATP and thus is independent of topo IIs catalytic activity. We also found that condensation and catalysis can concurrently proceed with minimal mutual interference. Our findings suggest topo II may directly participate in chromosome condensation during mitosis.

biophysics↗

bioRxiv2024

Torsion is a Dynamic Regulator of DNA Replication Stalling and Reactivation

The inherent helical structure of DNA dictates that a replisome must rotate relative to DNA during replication, presenting inevitable topological challenges to replication. However, little is known about how the replisome progresses against torsional stress. Here, we developed a label-free, high-resolution, real-time assay to monitor replisome movement under torsion. We visualized the replisome rotation of DNA and determined how the replisome slows down under torsion. We found that while helicase or DNA polymerase (DNAP) individually is a weak torsional motor, the replisome composed of both enzymes is the most powerful DNA torsional motor studied to date. It generates [~] 22 pN{middle dot}nm of torque before stalling, twice the stall torque of E. coli RNA polymerase. Upon replisome stalling, the specific interaction between helicase and DNAP stabilizes the fork junction; without it, the fork can regress hundreds of base pairs. We also discovered that prolonged torsion-induced stalling inactivates the replisome. Surprisingly, DNAP exchange, mediated by the helicase, is highly effective in facilitating replication restart, but only if excess DNAP is present during stalling. Thus, helicase and DNA polymerase work synergistically as a powerful torsional motor, and their dynamic and fluid interactions are crucial for maintaining fork integrity under torsional stress. This work demonstrates that torsion is a strong regulator of DNA replication stalling and reactivation.

biophysics↗

bioRxiv2024

RNA Polymerase II is a Polar Roadblock to a Progressing DNA Fork

DNA replication and transcription occur simultaneously on the same DNA template, leading to inevitable conflicts between the replisome and RNA polymerase. These conflicts can stall the replication fork and threaten genome stability. Although numerous studies show that head-on conflicts are more detrimental and more prone to promoting R-loop formation than co-directional conflicts, the fundamental cause for the RNA polymerase roadblock polarity remains unclear, and the structure of these R-loops is speculative. In this work, we use a simple model system to address this complex question by examining the Pol II roadblock to a DNA fork advanced via mechanical unzipping to mimic the replisome progression. We found that the Pol II binds more stably to resist removal in the head-on configuration, even with minimal transcript size, demonstrating that the Pol II roadblock has an inherent polarity. However, an elongating Pol II with a long RNA transcript becomes an even more potent and persistent roadblock while retaining the polarity, and the formation of an RNA-DNA hybrid mediates this enhancement. Surprisingly, we discovered that when a Pol II collides with the DNA fork head-on and becomes backtracked, an RNA-DNA hybrid can form on the lagging strand in front of Pol II, creating a topological lock that traps Pol II at the fork. TFIIS facilitates RNA-DNA hybrid removal by severing the connection of Pol II with the hybrid. We further demonstrate that this RNA-DNA hybrid can prime lagging strand replication by T7 DNA polymerase while Pol II is still bound to DNA. Our findings capture basal properties of the interactions of Pol II with a DNA fork, revealing significant implications for transcription-replication conflicts.

biophysics↗

bioRxiv2024

High-Resolution Genome-Wide Mapping of Chromatin Accessibility and Torsional Stress

Torsional stress in chromatin plays a fundamental role in cellular functions, influencing key processes such as transcription, replication, and chromatin organization. Transcription and other processes may generate and be regulated by torsional stress. In the genome, the interplay of these processes creates complicated patterns of both positive (+) and negative (-) torsion. However, a challenge in generating an accurate torsion map is determining the zero-torsion baseline signal, which is conflated with chromatin accessibility. Here, we introduce a high-resolution method based on the intercalator trimethylpsoralen (TMP) to address this challenge. We describe a method to establish the zero-torsion baseline while preserving the chromatin state of the genome of S. cerevisiae. This approach enables both high-resolution mapping of accessibility and torsional stress in chromatin in the cell. Our analysis shows transcription-generated torsional domains consistent with the twin-supercoiled-domain model of transcription and suggests a role for torsional stress in recruiting topoisomerases and in regulating 3D genome architecture via cohesin. Significantly, we reveal that insulator sequence-specific transcription factors decouple torsion between divergent promoters, whereas torsion spreads between divergent promoters lacking these factors, suggesting that torsion serves as a regulatory mechanism in these regions. Although insulators are known to decouple gene expression, our finding provides a physical explanation of how such decoupling may occur. This new method provides a potential path forward for using TMP to measure torsional stress in the genome without the confounding contribution of accessibility in chromatin.

genomics↗

bioRxiv2024

Geometry of Braided DNA Dictates Supercoiling Partition

DNA's helical structure requires the replisome to rotate relative to the parental DNA during replication, generating supercoiling that partitions ahead and behind the fork. The influence of DNA substrate geometry on supercoiling partitioning and torsional resistance remains unclear. Here, we have engineered DNA-braiding substrates with end separations found during replication, measured braiding torques using an angular optical trap, and interpreted the results using Monte Carlo simulations. A substrate with small separations readily initiates braiding, suggesting that the fork would rotate during replication, partitioning supercoiling behind the replisome and reducing torsional resistance. In contrast, a substrate with a substantial separation at one end imposes a pronounced "torsional barrier" that hinders braiding initiation, suggesting a lack of fork rotation during replication, so supercoiling would partition ahead of the replisome. Our findings reveal a physical mechanism in which DNA fork geometry could modulate replisome dynamics, manage topological stress, and regulate replication progression in vivo.

biophysics↗

bioRxiv2024

Torsional Mechanics of Circular DNA

Circular DNA found in the cell is actively regulated to an underwound state, with their superhelical density close to{sigma} [~] - 0.06. While this underwound state is essential to life, how it impacts the torsional mechanical properties of DNA is not fully understood. In this work, we performed simulations to understand the torsional mechanics of circular DNA and validated our results with single-molecule measurements and analytical theory. We found that the torque generated at{sigma} [~] - 0.06 is near but slightly below that required to melt DNA, significantly decreasing the energy barrier for proteins that interact with melted DNA. Furthermore, supercoiled circular DNA experiences force (tension) and torque that are equally distributed through the DNA contour. We have also extended a previous analytical framework to show how the plectonemic twist persistence length depends on the intrinsic bending persistence length and twist persistence length. Our work establishes a framework for understanding DNA supercoiling and torsional dynamics of circular DNA.

biophysics↗

bioRxiv2024

DNA Polymerase Locks Replication Fork Under Stress

Replication of DNA requires the parental DNA to be unwound to allow the genetic information to be faithfully duplicated by the replisome. While this function is usually shared by a host of proteins in the replisome, notably DNA polymerase (DNAP) and helicase, the consequence of DNAP synthesizing DNA while decoupled from helicase remains not well understood. The unwinding of downstream DNA poses significant stress to DNAP, and the interaction between DNAP and the replication fork may affect replication restart. In this work, we examined the consequences of DNAP working against the stress of the DNA replication fork. We found that prolonged exposure of DNAP to the stress of the replication fork inactivates replication. Surprisingly, replication inactivation was often accompanied by a strong DNAP interaction with the leading and lagging strands at the fork, locking the fork in place. We demonstrated that fork locking is a consequence of DNAP forward translocation, and the exonuclease activity of DNAP, which allows DNAP to move in reverse, is essential in protecting the fork from inactivation. Furthermore, we found the locking configuration is not reversible by the subsequent addition of helicase. Collectively, this study provides a deeper understanding of the DNAP-fork interaction and mechanism in keeping the replication fork active during replication stress.

biophysics↗

bioRxiv2024

Tunable Elliptical Cylinders for Rotational Mechanical Studies of Single DNA Molecules

The angular optical trap (AOT) is a powerful technique for measuring the DNA topology and rotational mechanics of fundamental biological processes. Realizing the full potential of the AOT requires rapid torsional control of these processes. However, existing AOT quartz cylinders are limited in their ability to meet the high rotation rate requirement while minimizing laser-induced photodamage. In this work, we present a novel trapping particle design to meet this challenge by creating small metamaterial elliptical cylinders with tunable trapping force and torque properties. The optical torque of these cylinders arises from their shape anisotropy, with their optical properties tuned via multilayered SiO2 and Si3N4 deposition. We demonstrate that these cylinders can be rotated at about 3 times the rate of quartz cylinders without slippage while enhancing the torque measurement resolution during DNA torsional elasticity studies. This approach opens new opportunities for previously inaccessible rotational studies of DNA processing.

biophysics↗

bioRxiv2024

An Effective Surface Passivation Assay for Single-Molecule Studies of Chromatin and Topoisomerase II

For single-molecule studies requiring surface anchoring of biomolecules, a poorly passivated surface can result in alterations of biomolecule structure and function that can result in artifacts. This protocol describes surface passivation and sample chamber preparation for mechanical manipulation of chromatin fibers and characterization of topoisomerase II activity in physiological buffer conditions. The method employs enhanced surface hydrophobicity and purified blocking proteins to reduce non-specific surface adsorption. This method is accessible, cost-effective, and potentially widely applicable to other biomolecules. For a complete list of publications that employ this protocol, see the paper references. B. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=154 SRC="FIGDIR/small/614989v1_ufig1.gif" ALT="Figure 1"> View larger version (32K): org.highwire.dtl.DTLVardef@d52b73org.highwire.dtl.DTLVardef@117bfc3org.highwire.dtl.DTLVardef@2eb670org.highwire.dtl.DTLVardef@cb88b2_HPS_FORMAT_FIGEXP M_FIG C_FIG

biophysics↗

bioRxiv2024

Optical Torque Calculations and Measurements for DNA Torsional Studies

The angular optical trap (AOT) is a powerful instrument for measuring the torsional and rotational properties of a biological molecule. Thus far, AOT studies of DNA torsional mechanics have been carried out using a high numerical aperture oil-immersion objective, which permits strong trapping, but inevitably introduces spherical aberrations due to the glass-aqueous interface. However, the impact of these aberrations on torque measurements is not fully understood experimentally, partly due to a lack of theoretical guidance. Here, we present a numerical platform based on the finite element method to calculate forces and torques on a trapped quartz cylinder. We have also developed a new experimental method to accurately determine the shift in the trapping position due to the spherical aberrations by using a DNA molecule as a distance ruler. We found that the calculated and measured focal shift ratios are in good agreement. We further determined how the angular trap stiffness depends on the trap height and the cylinder displacement from the trap center and found full agreement between predictions and measurements. As further verification of the methodology, we showed that DNA torsional properties, which are intrinsic to DNA, could be determined robustly under different trap heights and cylinder displacements. Thus, this work has laid both a theoretical and experimental framework that can be readily extended to investigate the trapping forces and torques exerted on particles with arbitrary shapes and optical properties. SIGNIFICANCEWe developed a simulation platform based on the finite element method for force and torque calculation for particles in an angular optical trap (AOT), with considerations of tightly focused Gaussian beam, spherical aberrations, and optically anisotropic particles. Experimental measurements of focal shift ratio, force, and torque under multiple conditions were in good agreement with predictions from the simulations. We also demonstrated that intrinsic DNA torsional properties can be robustly measured under different AOT measurement conditions, strongly validating our simulations and calibrations. Our platform can facilitate trapping particle design for single-molecule assays using the AOT.

biophysics↗

bioRxiv2023

Chromatinization Modulates Topoisomerase II Processivity

Type IIA topoisomerases are essential DNA processing enzymes that must robustly and reliably relax DNA torsional stress in vivo. While cellular processes constantly create different degrees of torsional stress, how this stress feeds back to control type IIA topoisomerase function remains obscure. Using a suite of single-molecule approaches, we examined the torsional impact on supercoiling relaxation of both naked DNA and chromatin by eukaryotic topoisomerase II (topo II). We observed that topo II was at least [~] 50-fold more processive on plectonemic DNA than previously estimated, capable of relaxing > 6000 turns. We further discovered that topo II could relax supercoiled DNA prior to plectoneme formation, but with a [~]100-fold reduction in processivity; strikingly, the relaxation rate in this regime decreased with diminishing torsion in a manner consistent with the capture of transient DNA loops by topo II. Chromatinization preserved the high processivity of the enzyme under high torsional stress. Interestingly, topo II was still highly processive ([~] 1000 turns) even under low torsional stress, consistent with the predisposition of chromatin to readily form DNA crossings. This work establishes that chromatin is a major stimulant of topo II function, capable of enhancing function even under low torsional stress.

biophysics↗

bioRxiv2022

Novel Methods for Elucidating Modality Importance in Multimodal Electrophysiology Classifiers

Multimodal classification is increasingly common in biomedical informatics studies. Many such studies use deep learning classifiers with raw data, which makes explainability difficult. As such, only a few studies have applied explainability methods, and new methods are needed. In this study, we propose sleep stage classification as a testbed for method development and train a convolutional neural network with electroencephalogram (EEG), electrooculogram, and electromyogram data. We then present a global approach that is uniquely adapted for electrophysiology analysis. We further present two local approaches that can identify subject-level differences in explanations that would be obscured by global methods and that can provide insight into the effects of clinical and demographic variables upon the patterns learned by the classifier. We find that EEG is globally the most important modality for all sleep stages, except non-rapid eye movement stage 1 and that local subject-level differences in importance arise. We further show that sex, followed by medication and age had significant effects upon the patterns learned by the classifier. Our novel methods enhance explainability for the growing field of multimodal classification, provide avenues for the advancement of personalized medicine, and yield novel insights into the effects of demographic and clinical variables upon classifiers.

bioinformatics↗

bioRxiv2021

A Novel Local Ablation Approach for Explaining Multimodal Classifiers

With the growing use of multimodal data for deep learning classification in healthcare research, more studies have begun to present explainability methods for insight into multimodal classifiers. Among these studies, few have utilized local explainability methods, which could provide (1) insight into the classification of each sample and (2) an opportunity to better understand the effects of demographic and clinical variables within datasets (e.g., medication of subjects in electrophysiology data). To the best of our knowledge, this opportunity has not yet been explored within multimodal classification. We present a novel local ablation approach that shows the importance of each modality to the correct classification of each class and explore the effects of demographic and clinical variables upon the classifier. As a use-case, we train a convolutional neural network for automated sleep staging with electroencephalogram (EEG), electrooculogram (EOG), and electromyogram (EMG) data. We find that EEG is the most important modality across most stages, though EOG is particular important for non-rapid eye movement stage 1. Further, we identify significant relationships between the local explanations and subject age, sex, and state of medication which suggest that the classifier learned specific features associated with these variables across multiple modalities and correctly classified samples. Our novel explainability approach has implications for many fields involving multimodal classification. Moreover, our examination of the degree to which demographic and clinical variables may affect classifiers could provide direction for future studies in automated biomarker discovery.

bioinformatics↗

bioRxiv2021

An Explainable Deep Learning Approach for Multimodal Electrophysiology Classification

In recent years, more biomedical studies have begun to use multimodal data to improve model performance. As such, there is a need for improved multimodal explainability methods. Many studies involving multimodal explainability have used ablation approaches. Ablation requires the modification of input data, which may create out-of-distribution samples and may not always offer a correct explanation. We propose using an alternative gradient-based feature attribution approach, called layer-wise relevance propagation (LRP), to help explain multimodal models. To demonstrate the feasibility of the approach, we selected automated sleep stage classification as our use-case and trained a 1-D convolutional neural network (CNN) with electroencephalogram (EEG), electrooculogram (EOG), and electromyogram (EMG) data. We applied LRP to explain the relative importance of each modality to the classification of different sleep stages. Our results showed that across all samples, EEG was most important, followed by EOG, and EMG. For individual sleep stages, EEG and EOG had higher relevance for classifying awake and non-rapid eye movement 1 (NREM1). EOG was most important for classifying REM, and EEG was most relevant for classifying NREM2-NREM3. Also, LRP gave consistent levels of importance to each modality for correctly classified samples across folds and inconsistent levels of importance for incorrectly classified samples. Our results demonstrate the additional insight that gradient-based approaches can provide relative to ablation methods and highlight their feasibility for explaining multimodal electrophysiology classifiers.

bioinformatics↗

bioRxiv2021

Explainable Sleep Stage Classification with Multimodal Electrophysiology Time-series

Many automated sleep staging studies have used deep learning approaches, and a growing number have used multimodal data to improve their classification performance. However, few studies using multimodal data have provided model explainability. Some have used traditional ablation approaches that "zero out" a modality. However, the samples that result from this ablation are unlikely to be found in real electroencephalography (EEG) data, which could adversely affect the importance estimates that result. Here, we train a convolutional neural network for sleep stage classification with EEG, electrooculograms (EOG), and electromyograms (EMG) and propose an ablation approach that replaces each modality with values that approximate the line-related noise commonly found in electrophysiology data. The relative importance that we identify for each modality is consistent with sleep staging guidelines, with EEG being important for most sleep stages and EOG being important for Rapid Eye Movement (REM) and nonREM stages. EMG showed low relative importance across classes. A comparison of our approach with a "zero out" ablation approach indicates that while the importance results are consistent for the most part, our method accentuates the importance of modalities to the model for the classification of some stages like REM (p < 0.05). These results suggest that a careful, domain-specific selection of an ablation approach may provide a clearer indicator of modality importance. Further, this study provides guidance for future research on using explainability methods with multimodal electrophysiology data.

bioinformatics↗