Biology subjects

Baek, S.

Publications and source records attributed to Baek, S..

35 records · Page 2Linked to original sources

bioRxiv2022

Invariance of object detection in untrained deep neural networks

The ability to perceive visual objects with various types of transformations, such as rotation, translation, and scaling, is crucial for consistent object recognition. In machine learning, invariant object detection for a network is often implemented by augmentation with a massive number of training images, but the mechanism of invariant object detection in biological brains -- how invariance arises initially and whether it requires visual experience -- remains elusive. Here, using a model neural network of the hierarchical visual pathway of the brain, we show that invariance of object detection can emerge spontaneously in the complete absence of learning. First, we found that units selective to a particular object class arise in randomly initialized networks even before visual training. Intriguingly, these units show robust tuning to images of each object class under a wide range of image transformation types, such as viewpoint rotation. We confirmed that this "innate" invariance of object selectivity enables untrained networks to perform an object-detection task robustly, even with images that have been significantly modulated. Our computational model predicts that invariant object tuning originates from combinations of non-invariant units via random feedforward projections, and we confirmed that the predicted profile of feedforward projections is observed in untrained networks. Our results suggest that invariance of object detection is an innate characteristic that can emerge spontaneously in random feedforward networks. HighlightsO_LIObject-selective units spontaneously emerge in untrained deep neural networks C_LIO_LIObject selectivity maintains robustly in a wide range of image transformations C_LIO_LIFeedforward model can explain spontaneous emergence of the invariance C_LIO_LIInnate invariance enables invariant object detection without learning to variations C_LI

neuroscience↗

bioRxiv2022

Combination of an inject-and-transfer system for serial femtosecond crystallography

Serial femtosecond crystallography (SFX) enables the determination of the room-temperature crystal structure of macromolecules without causing radiation damage and provides time-resolved molecular dynamics by pump-probe experiments. In the SFX experiment, the injector-based sample delivery method continuously provides fresh crystals to X-rays, and the fixed-target scanning method can be programmed to move the crystals to the desired location. This study introduces a combination of the inject-and-transfer system (BITS) method for sample delivery for SFX experiments, a hybrid injection, and a fixed-target scanning method. BITS allows solution samples to be reliably deposited on an ultraviolet ozone (UVO)-treaed polyimide films at flow rates as low as 1 nl/min. In application of BITS in SFX experiment, the lysozyme crystal samples were embedded in a viscous lard medium and injected at a 50-100 nl/min flow rate through a syringe needle onto an UVO-treated polyimide film mounted on a fixed-target scan stage. The deposited crystal sample on film were raster scanned to XFEL by motion stage in the horizontal and vertical directions. Using this method, we successfully determined the room-temperature structure of lysozyme at 2.1 [A] resolution. This method can be applied to the SFX experiments.

biophysics↗

bioRxiv2022

Single-cell Transcriptome and Accessible Chromatin Dynamics During Endocrine Pancreas Development

Delineating gene regulatory networks that orchestrate cell-type specification is an ongoing challenge for developmental biology studies. Single-cell analyses offer opportunities to address these challenges and accelerate discovery of rare cell lineage relationships and mechanisms underlying hierarchical lineage decisions. Here, we describe the molecular analysis of pancreatic endocrine cell differentiation using single-cell gene expression, chromatin accessibility assays coupled to genetic labeling and cell sorting. We uncover transcription factor networks that delineate {beta}-, - and{delta} -cell lineages. Through genomic footprint analysis we identify transcription factor-regulatory DNA interactions governing pancreatic cell development at unprecedented resolution. Our analysis suggests that the transcription factor Neurog3 may act as a pioneer transcription factor to specify the pancreatic endocrine lineage. These findings could improve protocols to generate replacement endocrine cells from renewable sources, like stem cells, for diabetes therapy.

developmental biology↗

bioRxiv2021

High-resolution single cell transcriptome analysis of zebrafish sensory hair cell regeneration

Loss of sensory hair cells in the mammalian inner ear leads to permanent hearing and vestibular defects, whereas loss of hair cells in zebrafish results in their regeneration. We used scRNA-Seq to characterize the transcriptional dynamics of hair cell regeneration in zebrafish at unprecedented spatio-temporal resolution. We uncovered three, sequentially activated modules. First, an injury/inflammatory response and downregulation of progenitor/stem cell maintenance genes within minutes after hair cell loss. Second, the transient activation of regeneration-specific genes. And third, a robust reactivation of developmental gene programs, including hair cell specification, cell cycle activation, ribosome biogenesis, and a metabolic switch to oxidative phosphorylation. The results are not only relevant for our understanding of hair cell regeneration and how we might be able to trigger it in mammals but also for regenerative processes in general. The data is searchable and publicly accessible via a web-based interface.

developmental biology↗

bioRxiv2021

Attrition Rate in Infant fNIRS Research: A Meta-Analysis

Understanding the trends and predictors of attrition rate, or the proportion of collected data that is excluded from the final analyses, is important for accurate research planning, assessing data integrity, and ensuring generalizability. In this pre-registered meta-analysis, we reviewed 182 publications in infant (0-24 months) functional near-infrared spectroscopy (fNIRS) research published from 1998 to April 9, 2020 and investigated the trends and predictors of attrition. The average attrition rate was 34.23% among 272 experiments across all 182 publications. Among a subset of 136 experiments which reported the specific reasons of subject exclusion, 21.50% of the attrition were infant-driven while 14.21% were signal-driven. Subject characteristics (e.g., age) and study design (e.g., fNIRS cap configuration, block/trial design, and stimulus type) predicted the total and subject-driven attrition rates, suggesting that modifying the recruitment pool or the study design can meaningfully reduce the attrition rate in infant fNIRS research. Based on the findings, we established guidelines on reporting the attrition rate for scientific transparency and made recommendations to minimize the attrition rates. We also launched an attrition rate calculator (LINK) to aid with research planning. This research can facilitate developmental cognitive neuroscientists in their quest toward increasingly rigorous and representative research. HighlightsO_LIAverage attrition rate in infant fNIRS research is 34.23% C_LIO_LI21.50% of the attrition are infant-driven (e.g., inattentiveness) while 14.21% are signal-driven (e.g., poor optical contact) C_LIO_LISubject characteristics (e.g., age) and study design (e.g., fNIRS cap configuration, block/trial design, and stimulus type) predict the total and infant-driven attrition rates C_LIO_LIModifying the recruitment pool or the study design can meaningfully reduce the attrition rate in infant fNIRS research C_LI

neuroscience↗

bioRxiv2021

Muscle fatigue revisited - Insights from optically pumped magnetometers

Muscle fatigue is well characterized electromyographically, nevertheless only information about summed potential differences is detectable. In contrast, recently developed quantum sensors optically pumped magnetometers (OPMs) offer the advantage of recording both the electrical current propagation in the muscle as well as its geometry, by measuring the magnetic field generated by the muscular action potentials. Magnetomyographic investigation of muscle fatigue is still lacking and it is an open question whether fatigue is characterized similarly in magnetomyography (MMG) compared to electromyography (EMG). Herein, we investigated the muscle fatigue during a 3x1-min strong isometric contraction of the rectus femoris muscle of 12 healthy subjects using simultaneous EMG-MMG (4-channel surface EMG and 4 OPM along the rectus femoris muscle). Both EMG and MMG showed the characteristic frequency decrease in the signal magnitude during isometric contraction, which is typical for muscle fatigue. In addition, it was shown that the main part of this frequency decrease seems to occur in the circular component of the magnetic field around the muscle fibers and less longitudinally along the muscle fibers. Overall, these results show not only that magnetomyography is capable of reproducing the electromyographic standards in identifying muscular fatigue, but it also adds relevant information about the spatial characterization of the signal. Therefore, OPM-MMG offers new insights for the study of muscular activity and might serve as a new, supplementary neurophysiological method.

physiology↗

bioRxiv2021

Pkd1 and Wnt5a genetically interact to control lymphatic vascular morphogenesis in mice

Lymphatic vascular development is regulated by well-characterised signalling and transcriptional pathways. These pathways regulate lymphatic endothelial cell (LEC) migration, motility, polarity and and morphogenesis. Canonical and non-canonical WNT signalling pathways are known to control LEC polarity and development of lymphatic vessels and valves. PKD1, encoding Polycystin-1, is the most commonly mutated gene in polycystic kidney disease but has also been shown to be essential in lymphatic vascular morphogenesis. The mechanism by which Pkd1 acts during lymphangiogenesis remains unclear. Here we find that loss of non-canonical WNT signalling components Wnt5a and Ryk phenocopy lymphatic defects seen in Pkd1 knockout mice. To investigate genetic interaction, we generated Pkd1/Wnt5a double knockout mice. Loss of Wnt5a suppressed phenotypes seen in the lymphatic vasculature of Pkd1-/- mice and Pkd1 deletion suppressed phenotypes observed in Wnt5a-/- mice. Thus, we report mutually suppressive roles for Pkd1 and Wnt5a, with developing lymphatic networks restored to a more wild-type state in double mutant mice. This genetic interaction between Pkd1 and the non-canonical WNT signalling pathway ultimately controls LEC polarity and the morphogenesis of developing vessel networks. Our work suggests that Pkd1 acts at least in part by regulating non-canonical WNT signalling during the formation of lymphatic vascular networks.

developmental biology↗

bioRxiv2020

CRIMSON: An Open-Source Software Framework for Cardiovascular Integrated Modelling and Simulation

In this work, we describe the CRIMSON (CardiovasculaR Integrated Modelling and SimulatiON) software environment. CRIMSON provides a powerful, customizable and user-friendly system for performing three-dimensional and reduced-order computational haemodynamics studies via a pipeline which involves: 1) segmenting vascular structures from medical images; 2) constructing analytic arterial and venous geometric models; 3) performing finite element mesh generation; 4) designing, and 5) applying boundary conditions; 6) running incompressible Navier-Stokes simulations of blood flow with fluid-structure interaction capabilities; and 7) post-processing and visualizing the results, including velocity, pressure and wall shear stress fields. A key aim of CRIMSON is to create a software environment that makes powerful computational haemodynamics tools accessible to a wide audience, including clinicians and students, both within our research laboratories and throughout the community. The overall philosophy is to leverage best-in-class open source standards for medical image processing, parallel flow computation, geometric solid modelling, data assimilation, and mesh generation. It is actively used by researchers in Europe, North and South America, Asia, and Australia. It has been applied to numerous clinical problems; we illustrate applications of CRIMSON to real-world problems using examples ranging from pre-operative surgical planning to medical device design optimization. CRIMSON binaries for Microsoft Windows 10, documentation and example input files are freely available for download from www.crimson.software, and the source code with compilation instructions is available on GitHub https://github.com/carthurs/CRIMSONFlowsolver (CRIMSON Flowsolver) under the GPL v3.0 license, and https://github.com/carthurs/CRIMSONGUI (CRIMSON GUI), under the AGPL v3.0 license. Support is available on the CRIMSON Google Groups forum, located at https://groups.google.com/forum/#!forum/crimson-users.

bioengineering↗

bioRxiv2020

A Constrained Mixture Theory Model to Study Autoregulation in the Coronary Circulation

Coronary autoregulation is a short-term response manifested by a relatively constant flow over a wide range of perfusion pressures for a given metabolic state. This phenomenon is thought to be facilitated through a combination of mechanisms, including myogenic, shear dependent, and metabolic controls. The study of coronary autoregulation is challenging due to the coupled nature of the mechanisms and their differential effects through the coronary tree. In this paper, we developed a novel framework to study coronary autoregulation based on the constrained mixture theory. This structurally-motivated autoregulation model required calibration of anatomical and structural parameters of coronary trees via a homeostatic optimization approach using extensive literature data. Autoregulation was then simulated for two different coronary trees: subepicardial and subendocardial. The structurally calibrated model reproduced available baseline hemodynamics and autoregulation data for each coronary tree. The autoregulation analysis showed that the diameter of the intermediate and small arterioles varies the most in response to changes in perfusion pressure. Finally, we demonstrated the utility of the model in two application examples: 1) response to drops in epicardial pressure, and 2) response to drug infusion in the coronary arteries. The proposed structurally-motivated model could be extended to study long-term growth and remodeling in the coronary circulation in response to hypertension, atherosclerosis, etc. Key pointsO_LICoronary autoregulation is defined as the capability of the coronary circulation to maintain the blood supply to the heart over a range of perfusion pressures. This phenomenon is facilitated through intrinsic mechanisms that control the vascular resistance by regulating the mechanical function of smooth muscle cells. Understanding the mechanisms involved in coronary autoregulation is one of the most fundamental questions in coronary physiology. C_LIO_LIThis paper presents a structurally-motivated coronary autoregulation model that uses a nonlinear continuum mechanics approach to account for the morphometry and vessel wall composition in two coronary trees in the subepicardial and subendocardial layers. C_LIO_LIThe model is calibrated against diverse experimental data from literature and is used to study heterogeneous autoregulatory response in the coronary trees. This model drastically differs from previous models, which relied on lumped parameter model formulations, and is suited to the study of long-term pathophysiological growth and remodeling phenomena in coronary vessels. C_LI

physiology↗

bioRxiv2020

Optimized residue analysis method for broflanilide and its metabolites in agricultural produce using the QuEChERS method and LC-MS/MS

Since broflanilide is a newly developed pesticide, analytical methods are required to determine the corresponding pesticide residues in diverse crops and foods. In this study, a pesticide residue analysis method was optimized for the detection and quantification of broflanilide and its two metabolites, DM-8007 and S(PFH-OH)-8007, in brown rice, soybean, apple, green pepper, mandarin, and kimchi cabbage. Residue samples were extracted from the produce using QuEChERS acetate and citrate buffering methods and were purified by dispersive solid-phase extraction (d-SPE) using six different adsorbent compositions with varying amounts of primary secondary amine (PSA), C18, and graphitized carbon black. All the sample preparation methods gave low-to-medium matrix effects, as confirmed by liquid chromatography-tandem mass spectrometry using standard solutions and matrix-matched standards. In particular, the use of the citrate buffering method, in combination with purification by d-SPE using 25 mg of PSA and a mixture of other adsorbents, consistently gave low matrix effects that in the range from -18.3 to 18.8%. Pesticide recoveries within the valid recovery range 70-120% were obtained both with and without d-SPE purification using 25 mg of PSA and other adsorbents. Thus, the developed residue analysis method is viable for the determination of broflanilide and its metabolites in various crops.

plant biology↗

bioRxiv2020

Sparse long-range connections in visual cortex for cost-efficient small-world networks

The brain performs visual object recognition using much shallower hierarchical stages than artificial deep neural networks employ. However, the mechanism underlying this cost-efficient function is elusive. Here, we show that cortical long-range connectivity(LRC) may enable this parsimonious organization of circuits for balancing cost and performance. Using model network simulations based on data in tree shrews, we found that sparse LRCs, when added to local connections, organize a small-world network that dramatically enhances object recognition of shallow feedforward networks. We found that optimization of the ratio between LRCs and local connections maximizes the small-worldness and task performance of the network, by minimizing the total length of wiring needed for integration of the global information. We also found that the effect of LRCs varies by network size, which explains the existence of species-specific LRCs in mammalian visual cortex of various sizes. Our results demonstrate a biological strategy to achieve cost-efficient brain circuits. HighlightsO_LILong-range connections (LRCs) enhance the object recognition of shallow networks C_LIO_LISparse LRCs added to dense local connections organize a small-world type network C_LIO_LISmall-worldness of networks modulates the balance between performance and wiring cost C_LIO_LIDistinct LRCs in various species are due to the size-dependent effect of LRCs C_LI Significance statementThe hierarchical depth of the visual pathway in the brain is constrained by biological factors, whereas artificial deep neural networks consist of super-deep structures (i.e., as deep as computational power allows). Here, we show that long-range horizontal connections (LRCs) observed in mammalian visual cortex may enable shallow biological networks to perform cognitive tasks that require deeper artificial structures, by implementing cost-efficient organization of circuitry. Using model simulations based on anatomical data, we found that sparse LRCs, when added to dense local circuits, organize "small-world" type networks and that this dramatically enhances image classification performance by integrating both local and global components of visual stimulus. Our findings show a biological strategy of brain circuitry to balance sensory performance and wiring cost in the networks. One sentence summaryCortical long-range connections organize a small-world type network to achieve cost-efficient functional circuits under biological constraints

neuroscience↗

bioRxiv2020

Viscous medium-based crystal support in sample holder for fixed-target serial femtosecond crystallography

Serial femtosecond crystallography (SFX) enables the determination of a room-temperature crystal structure of macromolecules without causing radiation damage, as well as provides time-resolved molecular dynamics data in pump-probe experiments. Fixed-target SFX (FT-SFX) can minimize sample consumption and physical effects to crystals during sample delivery. Various types of sample holders have been developed and applied in FT-SFX; however, no sample holder has been developed that can universally mount crystals of various sizes and shapes. Here, we introduce a viscous media-based crystal support in a sample holder for FT-SFX. Crystal samples were embedded in viscous media such as gelatin and agarose, which were enclosed in a polyimide film. In the vertically placed sample holder, the viscous medium stably supported crystals between the two polyimide films without crystal sinking due to gravity. Using this method, we performed FT-SFX experiments with glucose isomerase and lysozyme embedded in gelatin and agarose, respectively. The room-temperature crystal structures of glucose isomerase and lysozyme were successfully determined at 1.75 and 1.80 [A] resolutions, respectively. Viscous media used in this experiment showed negligible background scattering in data processing. This method is useful for delivering crystal samples of various sizes and shapes in FT-SFX experiments.

biophysics↗

bioRxiv2019

Spontaneous generation of face recognition in untrained deep neural networks

Face-selective neurons are observed in the primate visual pathway and are considered the basis of facial recognition in the brain. However, it is debated whether this neuronal selectivity can arise spontaneously, or requires training from visual experience. Here, we show that face-selective neurons arise spontaneously in random feedforward networks in the absence of learning. Using biologically inspired deep neural networks, we found that face-selective neurons arise under three different network conditions: one trained using non-face natural images, one randomized after being trained, and one never trained. We confirmed that spontaneously emerged face-selective neurons show the biological view-point-invariant characteristics observed in monkeys. Such neurons suddenly vanished when feedforward weight variation declined to a certain level. Our results suggest that innate face-selectivity originates from statistical variation of the feedforward projections in hierarchical neural networks.

neuroscience↗

bioRxiv2019

Spontaneous generation of innate number sense in untrained deep neural networks

Number-selective neurons are observed in numerically naive animals, but it was not understood how this innate function emerges in the brain. Here, we show that neurons tuned to numbers can arise in random feedforward networks, even in the complete absence of learning. Using a biologically inspired deep neural network, we found that number tuning arises in three cases of networks: one trained to non-numerical natural images, one randomized after trained, and one never trained. Number-tuned neurons showed characteristics that were observed in the brain following the Weber-Fechner law. These neurons suddenly vanished when the feedforward weight variation decreased to a certain level. These results suggest that number tuning can develop from the statistical variation of bottom-up projections in the visual pathway, initializing innate number sense.

neuroscience↗

bioRxiv2019

Predicting potential current distribution of Lycorma delicatula (Hemiptera: Fulgoridae) using MaxEnt model in South Korea

Lycorma delicatula (Hemiptera: Fulgoridae) is invasive insect in Korea which causes plant damages by sucking and sooty molds. Lycorma delicatula was first detected in South Korea in 2004, where its introduction and spreading possibly were affected by human activity-related factors. Here, we used MaxEnt to describe current distribution of L. delicatula in Korea and tried to find out the impact of human influences for distribution. We used 143 sites of occurrence data, 19 bioclimatic variables, duration of temperature below -11{degrees}C, average daily minimum temperature in January, cumulative thermal unit variable, the distribution of grape orchard variable and human footprint to create models. These models were estimated by two sets of 24 candidates with feature combinations and regularization multipliers. In addition, these two sets were created as models with and without footprint for how human influence affect to distribution. Model selection for optimal model was performed by selecting a model with a lowest sum of each rank in small sample-size corrected Akaikes information criterion and difference between training and test AUC. Model of LQ10 parameter combinations was selected as optimal models for both model sets. Consequently, both of distribution maps from these models showed similar patterns of presence probability for L. delicatula. Both models expected that low altitude regions were relatively more suitable than mountain areas in Korea. Footprint might be limited for the distribution and L. delicatula might already occupy most of available habitats. Human-related factors might contribute to spread of L. delicatula to uninfected areas.

ecology↗

bioRxiv2018

Single cell RNA-Seq reveals distinct stem cell populations that drive sensory hair cell regeneration in response to loss of Fgf and Notch signaling

Loss of sensory hair cells leads to deafness and balance deficiencies. In contrast to mammalian hair cells, zebrafish ear and lateral line hair cells regenerate from poorly characterized, proliferating support cells. Equally ill-defined is the gene regulatory network underlying the progression of support cells to cycling hair cell progenitors and differentiated hair cells. We used single cell RNA-Sequencing (scRNA-Seq) of lateral line sensory organs and uncovered five different support cell types, including quiescent and activated stem cells. In silico ordering of support cells along a developmental trajectory identified cells that self-renew and new groups of genes required for hair cell differentiation. scRNA-Seq analyses of fgf3 mutants, in which hair cell regeneration is increased, demonstrates that Fgf and Notch signaling inhibit proliferation of support cells in parallel by inhibiting Wnt signaling. Our scRNA-Seq analyses set the foundation for mechanistic studies of sensory organ regeneration and is crucial for identifying factors to trigger hair cell production in mammals. As a resource, we implemented a shiny application that allows the community to interrogate cell type specific expression of genes of interest.

developmental biology↗

bioRxiv2018

Fast temporal dynamics and causal relevance of face processing in the human temporal cortex

Recordings with a large number of intracranial electrodes in eight neurosurgical subjects offered a unique opportunity to examine the fast temporal dynamics of face processing simultaneously across a relatively large extent of the human temporal cortex (TC). Measuring the power of slow oscillatory bands of activity ({theta}, , {beta}, and {gamma}) as well as High-Frequency Broadband (HFB, 70-177 Hz) signal, we found that the HFB showed the strongest univariate and multivariate changes in response to face compared to non-face stimuli. Using the HFB signal as a surrogate marker for local cortical engagement, we identified recording sites with selective responses to faces that were anatomically consistent across subjects and responded with graded strength to human, mammal, bird, and marine animal faces. Importantly, the most face selective sites were located more posteriorly and responded earlier than those with less selective responses to faces. Using machine learning based methods, we demonstrated that a sparse model focusing on information from the human face selective sites performed as well as, or better than, anatomically distributed models of face processing when discriminating faces from non-faces stimuli. Lastly, we identified the posterior fusiform (pFUS) site as causally the most relevant node for inducing distortion of face perception by direct electrical stimulation. Our findings support the notion of face information being processed first in the most selective sites - that are anatomically discrete and localizable within individual brains and anatomically consistent across subjects - which is then distributed in time to less selective anterior temporal sites within a time window that is too fast to be detected by current neuroimaging methods. The new information about the fast spatio-temporal dynamics of face processing across multiple sites of the human brain provides a new common ground for unifying the seemingly contradictory modular and distributed models of face processing in the human brain.

neuroscience↗