bioRxiv Science⌕ Search

Biology subjects

Ko, E.

Publications and source records attributed to Ko, E..

18 recordsLinked to original sources
bioRxiv2026

PINT: Pathway-pathway interactions for predicting interpretable clinical outcomes from gene expression

MotivationDisease mechanisms emerge from the coordinated activity of multiple biological pathways, rather than from individual pathways acting in isolation. Existing pathway-based deep learning models, however, treat pathways as independent entities, aggregating their representations through fully connected layers that disregard inter-pathway relationships. This architectural limitation overlooks an important dimension of disease biology, potentially constraining both predictive performance and the capacity to generate biologically meaningful interpretations. ResultsWe introduce a pathway-based attentive interpretability model, named PINT, that models interactions among pathways through a self-attention mechanism from gene expression data. An attention-based pooling layer further identifies patient-specific pathway contributions to the final prediction. Evaluation across five TCGA cancer datasets demonstrated that PINT consistently outperformed benchmark models in survival analysis. More importantly, PINT identifies pathways significantly associated with survival as well as reveals biologically meaningful interactions among pathways. In the BRCA dataset, PINT identified significant pathways, pathway-pathway interactions, and gene-level contributions within pathways for individual patients, most of which were supported by existing literature. Specifically, the RAS signaling pathway emerged as significantly associated with patient survival, and the learned interaction scores recovered known relationships between RAS signaling and several regulatory pathways, including cAMP, TNF, and Rap1 signaling. Availability and implementationThe source code and data are available at https://github.com/datax-lab/PINT.

bioinformatics↗
bioRxiv2026

Cortex-Wide Cellular Imaging in Freely Locomoting Mice Using Cortex Camera Array Microscope (CortexCAM)

Understanding single-cell neuronal activity across multiple brain regions in the context of ethologically relevant behaviors is a major goal in systems neuroscience. We have engineered the Cortex Camera Array Microscope (CortexCAM), integrating four miniaturized fluorescence imaging microscopes to simultaneously capture cellular activity from contiguous fields of view spanning over 48 mm2 of the dorsal cortex. The CortexCAM is capable of imaging > 9000 individual neurons across much of the primary and secondary motor, somatosensory, visual, retrosplenial, and association cortices across both hemispheres of the dorsal cortex. The compact nature of the CortexCAM allows integration into a passive mechanical gantry system to form the mobile CortexCAM. The mobile CortexCAM allows volitional control of the animals translational motion (x, y) and rotational motion (yaw) in physical behavior arenas. We then use the mobile CortexCAM to perform cortex-wide cellular resolution imaging in freely locomoting mice performing alternating choice tasks, as well as during social interactions. Thus the CortexCAM allows studying cortex-wide cellular dynamics in behaviors that cannot be achieved in headfixed settings.

neuroscience↗
bioRxiv2026

Fairness-aware Supervised Hierarchical Contrastive Semantic Learning for Sexual Dimorphism Analysis

MotivationSexual dimorphism is a fundamental biological determinant driving systematic differences in disease susceptibility, progression, and clinical outcomes. However, current AI-based genomic models often exhibit algorithmic bias and fail to capture these sex-specific mechanisms, creating a critical barrier to unbiased precision medicine. Ensuring fairness in the context of sexual dimorphism requires understanding and addressing the distinct biological mechanisms functioning in each sex, rather than focusing solely on equalizing predictive performance. ResultsWe propose a fairness-aware supervised hierarchical contrastive learning approach, called FairHICON, to discover unbiased sex-common and sex-specific genomic drivers. Evaluations on cancer and asthma transcriptomic datasets demonstrate that FairHICON significantly outperforms state-of-the-art benchmarks, improving predictive performance by up to 9% while effectively reducing the performance gap between male and female cohorts. Furthermore, prognostic validation confirms that the identified sex-specific pathways stratify patient survival significantly better within their corresponding sex groups. This validates FairHICON to elucidate the molecular heterogeneity of sexual dimorphism, advancing inclusive precision medicine. Availability and implementationThe source code and data is available at https://github.com/datax-lab/FairHICON.

bioinformatics↗
bioRxiv2026

PIMO: Pathway-based Interpretable Multi-Omics interactions for multi-omics integration

MotivationModeling inter-omics interactions across multiple molecular levels is critical for deciphering the mechanisms underlying complex diseases. Epigenomic and structural alterations, such as DNA methylation and copy number alterations, modulate gene expression and collectively influence disease progression and patient survival outcomes. Despite advancements in deep learning-based multi-omics analysis, gene-level interactions of inter-omics have been seldom considered, due to combinational complexity and power, which limits interpretability and mechanistic insight. ResultsWe propose a Pathway-based Interpretable deep learning Multi-Omics interaction model, PIMO, that explicitly captures regulatory effects across omics layers. Experiments on multiple TCGA cancer datasets showed that PIMO consistently outperformed state-of-the-art baselines in survival analysis, up to 13% increase in the C-index. PIMO provides biologically interpretable analyses that identify important pathways, genes, and inter-omics interactions with DNA methylation and copy number alterations. Availability and implementationThe source code and data is available at https://github.com/datax-lab/PIMO.

bioinformatics↗
bioRxiv2024

Computer vision guided rapid and precise automated cranial microsurgeries in rodents

Neuroscientists employ various experimental procedures to interface with the brain to study and perturb the neural activity during behavior. A common procedure that allows such physical interfacing is cranial microsurgery, wherein small to large craniotomies are performed in the overlying skull for insertion of neural interfaces or implantation of optically clear windows for long-term cranial observation. Performing craniotomies is, however, a skilled task that requires significant time and practice and further needs to be carried out precisely to ensure that the procedure does not cause damage to the underlying brain and dura. Here, we present a computer vision-guided craniotomy robot (CV-Craniobot) that utilizes machine learning to accurately estimate the dorsal skull anatomy from optical coherence tomography (OCT) images. Instantaneous information of the skull morphology is used by a robotic mill to rapidly and precisely remove the skull from a desired craniotomy location. We show that the CV-Craniobot can perform small (2 - 4 mm diameter) craniotomies with near 100% success rates within 2 minutes and large craniotomies encompassing most of the dorsal cortex in less than 5 minutes. Thus, the CV-Craniobot enables rapid and precise craniotomies, significantly reducing surgery time as compared to human practitioners and eliminating the need for long training.

neuroscience↗
bioRxiv2024

Dual-color micro-LEDs integrated neural interface for multi-control optogenetic electrophysiology

Over recent decades, optogenetics has become a pivotal technique for elucidating the functionality of neuronal circuits in living organisms. By genetically modifying specific cells within targeted tissues to respond to particular optical stimuli, researchers can achieve precise activation or inhibition of these cells. This capability enables detailed investigations of neural circuitry with unprecedented accuracy. However, there is a rising need for hardware that supports bidirectional control in conjunction with electrophysiological recording. A significant challenge in this domain is the compact integration of dual light sources and a recording system. This study addresses this challenge through the development of a novel microfabrication and assembly technique for embedding dual-color micro-LEDs and recording electrodes into a Michigan-type neural probe structure, designated as DuoLite (Dual-color micro-LEDs Integrated Neural-Interface Optoelectrode for Multi-Control Optogenetic Electrophysiology). We present two device variants: (a) a small-group and (b) a large-group cell-targeted design, each incorporating micro-LEDs with a minimal area of <100 m2 for both red and blue light. The design and assembly techniques for integrating all three components within a shank width of <100 m are thoroughly detailed, and the functionality of the devices is validated through in vivo experiments.

bioengineering↗
bioRxiv2024

Pan-cortical cellular imaging in freely behaving mice using a miniaturized micro-camera array microscope (mini-MCAM)

Understanding how circuits in the brain simultaneously coordinate their activity to mediate complex ethnologically relevant behaviors requires recording neural activities from distributed populations of neurons in freely behaving animals. Current miniaturized imaging microscopes are typically limited to imaging a relatively small field of view, precluding the measurement of neural activities across multiple brain regions. Here we present a miniaturized micro-camera array microscope (mini-MCAM) that consists of four fluorescence imaging micro-cameras, each capable of capturing neural activity across a 4.5 mm x 2.55 mm field of view (FOV). Cumulatively, the mini-MCAM images over 30 mm2 area of sparsely expressed GCaMP6s neurons distributed throughout the dorsal cortex, in regions including the primary and secondary motor, somatosensory, visual, retrosplenial, and association cortices across both hemispheres. We demonstrate cortex-wide cellular resolution in vivo Calcium (Ca2+) imaging using the mini-MCAM in both head-fixed and freely behaving mice.

neuroscience↗
bioRxiv2024

Self-Assembled Origami Neural Probes for Scalable, Multifunctional, Three-Dimensional Neural Interface

Flexible intracortical neural probes have drawn attention for their enhanced longevity in high-resolution neural recordings due to reduced tissue reaction. However, the conventional monolithic fabrication approach has met significant challenges in: (i) scaling the number of recording sites for electrophysiology; (ii) integrating of other physiological sensing and modulation; and (iii) configuring into three-dimensional (3D) shapes for multi-sided electrode arrays. We report an innovative self-assembly technology that allows for implementing flexible origami neural probes as an effective alternative to overcome these challenges. By using magnetic-field-assisted hybrid self-assembly, multiple probes with various modalities can be stacked on top of each other with precise alignment. Using this approach, we demonstrated a multifunctional device with scalable high-density recording sites, dopamine sensors and a temperature sensor integrated on a single flexible probe. Simultaneous large-scale, high-spatial-resolution electrophysiology was demonstrated along with local temperature sensing and dopamine concentration monitoring. A high-density 3D origami probe was assembled by wrapping planar probes around a thin fiber in a diameter of 80[~]105 m using optimal foldable design and capillary force. Directional optogenetic modulation could be achieved with illumination from the neuron-sized micro-LEDs (LEDs) integrated on the surface of 3D origami probes. We could identify angular heterogeneous single-unit signals and neural connectivity 360{degrees} surrounding the probe. The probe longevity was validated by chronic recordings of 64-channel stacked probes in behaving mice for up to 140 days. With the modular, customizable assembly technologies presented, we demonstrated a novel and highly flexible solution to accommodate multifunctional integration, channel scaling, and 3D array configuration.

neuroscience↗
bioRxiv2023

A compact, ultrahigh-density headstage with high-fidelity hybrid integration for large-scale deep-brain opto-electrophysiology

Recent neuroscientific research seeks to comprehend the sophisticated deep-brain networks of neural circuits consisting of large scale neuronal ensembles across multiple brain regions. An ideal way to unveil the complex connectome might be stimulating individual neurons with high spatial resolution in a broad range of brain, while seamlessly monitoring the correspondent neuronal activities. Optogenetics is known as a key technology to enable such a goal thanks to its high spatial and temporal selectivity in neuromodulation. Existing silicon probe technologies have been able to partially achieve such a goal by recording broad region of brain activities through multiple electrodes per shank, but those cannot complete perfect coverage due to the limited channel counts for the optogenetic stimulation. Here, we present an high-channel-count optogenetic system with simultaneous 256 recoding and 128 optogenetic stimulation sites, exhibiting the highest channel density ever reported, enabled by a flexible polyimide cable-based hybrid-integration of a low-stimulation-artifact micro-LED ({micro}LED) opto-electrode with a low-power and -noise, area-efficient CMOS interfacing integrated-circuit (IC). The presented optogenetic system provides 256-neuron-size electrodes (11 x 15 {micro}m2) with a 40 {micro}m inter-electrode pitch for high spatial oversampling in recording and 128-soma-size {micro}LEDs (8 x 11 {micro}m2) with a 20 {micro}m inter-LED pitch for single-cell resolution in stimulation, resulting in a vertical span of 640 {micro}m and a horizontal span of 2,100 {micro}m with a total 8 shanks. For versatility in optogenetics-based experiments from small rodents to primates with user-preferable settings, the system base that provides programmability of recording and stimulation parameters and rest of signal processing, such as filtering, digitization, and data transmission including serial peripheral interface (SPI) has also been designed within small area of 23.8 x 28.8 mm2 with only 3.5-gram weight, resulting in the highest channel density both in size (0.56 channels/mm2) and weight (109.71 channels/gram) among the state-of-the-art optogenetics-based neuromodulation systems. To verify the system operation in vivo, a compact optogenetics headstage has been also fabricated. Using the prepared optogenetic headstage, 169 isolated neurons have been observed with various stimulation intensities. The results offered in this article indicate that the presented hybrid integrated ultrahigh-density, high-channel-count headstage can be used to realize the massive-scale in-depth brain studies with optogenetics.

bioengineering↗
bioRxiv2023

Brain-wide neural recordings in mice navigating physical spaces enabled by a cranial exoskeleton

Complex behaviors are mediated by neural computations occurring throughout the brain. In recent years, tremendous progress has been made in developing technologies that can record neural activity at cellular resolution at multiple spatial and temporal scales. However, these technologies are primarily designed for studying the mammalian brain during head fixation - wherein the behavior of the animal is highly constrained. Miniaturized devices for studying neural activity in freely behaving animals are largely confined to recording from small brain regions owing to performance limitations. We present a cranial exoskeleton that assists mice in maneuvering neural recording headstages that are orders of magnitude larger and heavier than the mice, while they navigate physical behavioral environments. Force sensors embedded within the headstage are used to detect the mouses milli-Newton scale cranial forces which then control the x, y, and yaw motion of the exoskeleton via an admittance controller. We discovered optimal controller tuning parameters that enable mice to locomote at physiologically realistic velocities and accelerations while maintaining natural walking gait. Mice maneuvering headstages weighing up to 1.5 kg can make turns, navigate 2D arenas, and perform a navigational decision-making task with the same performance as when freely behaving. We designed an imaging headstage and an electrophysiology headstage for the cranial exoskeleton to record brain-wide neural activity in mice navigating 2D arenas. The imaging headstage enabled recordings of Ca2+ activity of 1000s of neurons distributed across the dorsal cortex. The electrophysiology headstage supported independent control of up to 4 silicon probes, enabling simultaneous recordings from 100s of neurons across multiple brain regions and multiple days. Cranial exoskeletons provide flexible platforms for largescale neural recording during the exploration of physical spaces, a critical new paradigm for unraveling the brain-wide neural mechanisms that control complex behavior.

neuroscience↗
bioRxiv2023

Distinct mesoscale cortical dynamics encode search strategies during spatial navigation

Spatial navigation is a complex cognitive process that involves neural computations in distributed regions of the brain. Little is known about how cortical regions are coordinated when animals navigate novel spatial environments or how that coordination changes as environments become familiar. We recorded mesoscale calcium (Ca2+) dynamics across large swathes of the dorsal cortex in mice solving the Barnes maze, a 2D spatial navigation task where mice used random, serial, and spatial search strategies to navigate to the goal. Cortical dynamics exhibited patterns of repeated calcium activity with rapid and abrupt shifts between cortical activation patterns at sub-second time scales. We used a clustering algorithm to decompose the spatial patterns of cortical calcium activity in a low dimensional state space, identifying 7 states, each corresponding to a distinct spatial pattern of cortical activation, sufficient to describe the cortical dynamics across all the mice. When mice used serial or spatial search strategies to navigate to the goal, the frontal regions of the cortex were reliably activated for prolonged durations of time (> 1s) shortly after trial initiation. These frontal cortex activation events coincided with mice approaching the edge of the maze from the center and were preceded by temporal sequences of cortical activation patterns that were distinct for serial and spatial search strategies. In serial search trials, frontal cortex activation events were preceded by activation of the posterior regions of the cortex followed by lateral activation of one hemisphere. In spatial search trials, frontal cortical events were preceded by activation of posterior regions of the cortex followed by broad activation of the lateral regions of the cortex. Our results delineated cortical components that differentiate goal- and non-goal oriented spatial navigation strategies.

neuroscience↗
bioRxiv2022

flexLiTE: flexible micro-LED integrated optoelectrodesfor minimally-invasive chronic deep-brain study

Understanding complex neuronal circuitry and its functions of a living organism requires a specialized tool which is capable of recording a large ensemble of neuronal signals at single cell resolution and modulating neuronal activities selectively in the target region of brains with high spatiotemporal resolution, while sustaining long-term chronic operation without significant tissue degeneration or device shifts. We hereby present an ultra-flexible, minimally-invasive, Michigan-type neural probe for chronic opto-electrophysiology studies: flexLiTE (flexible micro-LED integrated optoelectrodes). flexLiTE incorporates monolithically integrated, soma-sized inorganic micro-LEDs (12 individually operated) and 32 recording electrodes. Both stimulation and recording modalities were achieved by stacking two modules on a flexible substrate: one with micro-LEDs for neuromodulation and the other with recording sites, resulting in a 115 m-wide,12 m-thick, 10 mm-long optoelectrode. From prototype devices, we demonstrated the reliable operation of flexLiTEs for recording and modulation of hippocampal neurons in a freely moving mice for over [~]8 month.

bioengineering↗
bioRxiv2021

Empowering Engineered Muscle in Biohybrid Pump by Extending Connexin 43 Duration with Reduced Graphene Oxides

Engineered skeletal muscle act as therapeutics invaluable to treat injured or diseased muscle and a "living" material essential to assemble biological machinery. For normal development, skeletal myoblasts should express connexin 43, one of the gap junction proteins that promote myoblast fusion and myogenesis, during the early differentiation stage. However, myoblasts cultured in vitro often down-regulate connexin 43 before differentiation, limiting myogenesis and muscle contraction. This study demonstrates that tethering myoblasts with reduced graphene oxide (rGO) slows connexin 43 regression during early differentiation and increases myogenic mRNA synthesis. The whole RNA sequencing also confirms that the rGO on cells increases regulator genes for myogenesis, including troponin, while decreasing negative regulator genes. The resulting myotubes generated a three-fold larger contraction force than the rGO-free myotubes. Accordingly, a valveless biohybrid pump assembled with the rGO-tethered muscle increased the fluid velocity and flow rate considerably. The results of this study would provide an important foundation for developing physiologically relevant muscle and powering up biomachines that will be used for various bioscience studies and unexplored applications.

bioengineering↗
bioRxiv2021

Long- and short-range bimodal signals mediate mate location and recognition in yellow fever mosquitoes

Mate location and mate selection behavior in diurnally-active yellow fever mosquitoes, Aedes aegypti, take place in mating swarms but the mechanisms underlying swarm formation and long-range detection of females by males remain largely unexplored. In high-speed video recordings we show that incident light reflects off the wings of swarming males, and in behavioral experiments we demonstrate that swarm formation and mate recognition are mediated, in part, by these wingbeat light flashes and by wingbeat sound signals that operate at long and short range, respectively. To test for range-dependent effects of these signals, we presented mating swarms in form of two paired 8-LED assemblies that were fitted with micro-speakers and placed either well separated in a large space or side-by-side in a small space. In the large but not the small space, the LED assembly flashing light at the wingbeat frequency of females (665 Hz), and emitting their wingbeat sound (665 Hz), attracted and prompted 5.8-times more alightings by males than the LED assembly emitting constant light and wingbeat sound. In the small space, the LED assembly flashing light and emitting wingbeat sound induced 5.0-times more alightings by males than the LED assembly flashing light without wingbeat sound. The attractiveness of light flash signals to males increased with increasing numbers of signals but did not vary according to their wavelengths (UV or blue). Females responded to light flash signals of males. As predicted by the sensory drive theory, light flashes had no signal function for crepuscular house mosquitoes, Culex pipiens.

ecology↗
bioRxiv2020

Cheese or cheese infusions - Ecological traps for mosquitoes and spotted wing Drosophila

We tested the hypothesis that the "ecological trap" phenomenon (a mismatch between a habitats perceived attractiveness and its actual quality, resulting in a population sink) is exploitable for pest control. We selected mosquitoes as modal organisms, because selection of an oviposition site by adult female mosquitoes in response to its perceived attractiveness is of paramount importance for the development and survival of their larval offspring. In laboratory and/or field experiments, we show that (i) each of five cheese varieties tested (Raclette, Pecorino, Brie, Gruyere, Limburger) strongly attracts females of both the yellow fever mosquito, Aedes aegypti, and the common house mosquito, Cx. pipiens; (ii) cheese infusions, or headspace odorant extracts (HOEs) of cheese infusions, significantly affect oviposition choices by Cx. pipiens and Ae. aegypti, (iii) HOEs contain at least 13 odorants; (iv) in field settings, cheese infusions more effectively stimulate oviposition by Cx. pipiens and Culiseta incidens than bluegrass (Poa sp.) infusions, and also capture (by drowning) the spotted wing Drosophila (SWD); (v) the microbe composition of home-made cheese infusions modulates oviposition choices by mosquitoes; and (vi) the type of cheese infusion coupled with its nutritional content strongly affects the survivorship of mosquito larvae. In combination, our data show that microbial metabolites associated with cheese and cheese infusions are both attractive to adult mosquitoes seeking hosts and oviposition sites, respectively, and are toxic to mosquito larvae. These microbes and their metabolites could thus be coopted for both the attract and the kill function of "attract & kill" mosquito control tactics. Implementation of customizable and non-conventional nutritional media, such as home-made cheese infusions, as microbe-based ecological traps presents a promising concept which exploits insect ecology for insect control.

ecology↗
bioRxiv2018

Low-dose cadmium potentiates lung inflammatory response to 2009 pandemic H1N1 influenza virus in mice

BACKGROUNDCadmium (Cd) is a toxic, pro-inflammatory metal ubiquitous in the diet that accumulates in body organs due to inefficient elimination. Many individuals exposed to dietary Cd are also infected by seasonal influenza virus. The H1N1 strain causes mild to severe pneumonia which can be fatal.\n\nOBJECTIVESTo determine the influence of low-dose Cd exposure on inflammatory responses to H1N1 influenza A virus.\n\nMETHODSWe exposed mice to low-dose (1 mg CdCl2/l) Cd or vehicle (water) for 16 weeks prior to infection with a sub-lethal dose of H1N1. Lung inflammation was assessed by histopathology and flow cytometry. We used a combination of transcriptomics, metabolomics and bioinformatics to determine the molecular associations of inflammatory cells important in Cd-exacerbated responses.\n\nRESULTSCd-treated mice had increased lung tissue inflammatory cells, including neutrophils, monocytes, T lymphocytes and dendritic cells, following H1N1 infection. Lung genetic responses to infection (increasing TNF-a, interferon and complement, and decreasing myogenesis) were also exacerbated. Global correlations with immune cell counts, leading edge gene transcripts and metabolites revealed that Cd increased correlation of myeloid immune cells with pro-inflammatory genes, particularly interferon-{gamma}, and metabolites in amino acid, nucleobase, glycerophospholipid and vitamin B3 pathways.\n\nDISCUSSIONCd burden in mice increased inflammation in response to sub-lethal H1N1 challenge, which was coordinated by genetic and metabolic responses, and could provide new targets for intervention against lethal inflammatory pathology of clinical H1N1 infection.

pharmacology and toxicology↗
bioRxiv2018

SPORTS1.0: a tool for annotating and profiling non-coding RNAs optimized for rRNA- and tRNA- derived small RNAs

High-throughput RNA-seq has revolutionized the process of small RNA (sRNA) discovery, leading to a rapid expansion of sRNA categories. In addition to the previously well-characterized sRNAs such as microRNAs (miRNAs), Piwi-interacting RNA (piRNAs), and small nucleolar RNA (snoRNAs), recent emerging studies have spotlighted on tRNA-derived sRNAs (tsRNAs) and rRNA-derived sRNAs (rsRNAs) as new categories of sRNAs that bear versatile functions. Since existing software and pipelines for sRNA annotation are mostly focused on analyzing miRNAs or piRNAs, here we developed the sRNA annotation pipeline optimized for rRNA- and tRNA- derived sRNAs (SPORTS1.0). SPORTS1.0 is optimized for analyzing tsRNAs and rsRNAs from sRNA-seq data, in addition to its capacity to annotate canonical sRNAs such as miRNAs and piRNAs. Moreover, SPORTS1.0 can predict potential RNA modification sites based on nucleotide mismatches within sRNAs. SPORTS1.0 is precompiled to annotate sRNAs for a wide range of 68 species across bacteria, yeast, plant, and animal kingdoms, while additional species for analyses could be readily expanded upon end users input. For demonstration, by analyzing sRNA datasets using SPORTS1.0, we reveal that distinct signatures are present in tsRNAs and rsRNAs from different mouse cell types. We also find that compared to other sRNA species, tsRNAs bear the highest mismatch rate which is consistent with their highly modified nature. SPORTS1.0 is an open-source software and can be publically accessed at https://github.com/junchaoshi/sports1.0.

bioinformatics↗