SEARCH · Search PubMed

Results for “Deep generative model”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 73 records · Page 4Linked to original sources

PubMed1991

The influence of model parameter values on the prediction of skin surface temperature: II. Contact problems.

A model of heat transfer and temperature distribution in the skin and superficial tissues which is based on a finite difference numerical solution of the one-dimensional multilayer coupled bioheat equation is presented. The model is used to investigate the influence of the values chosen to represent the physiological and thermal properties of the tissues on the skin surface temperature after contact with an external medium. It was found that the skin blood flow and dermal conductivity were the main cutaneous parameters which influence the contact response, but in terms of normalized temperature the response was little influenced by cutaneous metabolic heat generation and deep dermal temperature. For contact with a good conductor, the transient behaviour was sensitive to the heat transfer coefficient on the outer surface and the thickness of the contact material, but insensitive to the conductivity of the material.

Humans↗

PubMed1995

Deep knowledge and computer-assisted instruction in cardiology.

In this paper, we describe a qualitative heart model that is part of a computing environment, CARDIOLAB, and whose role includes the diagnosis and Computer Assisted Instruction (CAI) in cardiology. The model is based on a "deep knowledge" approach to diagnosis. Deep knowledge representations model the inner works of complex physical systems. Explicit representations of system components, component functions, and behavior allow a principled form of reasoning that extends the classical rule-based, first-generation expert systems. One of the main advantage of model-based diagnosis resides in the possibility of providing explanations to observed facts or measured data. This feature can be incorporated into CAI programs with similar benefits.

Cardiology↗

PubMed2025

Caenorhabditis diversity on Pohnpei, Micronesia, provides evidence that the Elegans Supergroup has its roots in the Americas and diversified in the Pacific en route to Asia.

The microscopic nematode Caenorhabditis elegans stands unrivaled as a model for developmental biology, neurobiology, and genetics, but fundamental aspects of its ecology, biogeography, and natural history remain unknown. Leveraging recent findings that place its center of diversity in the cool, high-elevation forests of Hawaii, we performed an intensive survey of the Caenorhabditis fauna of Pohnpei, a high island in Micronesia that is home to the largest patch of high-elevation forest between Hawaii and East Asia. We found nine species of Caenorhabditis, five of them new, but not C. elegans. Most species were limited to the hot lowlands but three spanned the elevational range and one was found only in the cloudforest. Using the distribution of Caenorhabditis nematodes among habitat patches - individual rotting fruits or flowers - we parameterized simple models that capture key aspects of the population biology of these animals. We generated transcriptomes for the new species and inferred a phylogeny for 70 species of Caenorhabditis, based on 2955 genes. This phylogeny allowed us to perform the first quantitative biogeographic analysis for the group. Our analysis suggests that the deep ancestors of the Elegans Supergroup of species lived in the Americas, and that the Supergroup's subsequent diversification occurred in Remote Oceania. The ancestors of the Supergroup gave rise to a diverse Oceanian fauna and ultimately to multiple lineages that moved into Asia, Africa, Australasia, and back into the Americas. Though biogeographic inferences are limited by the lack of information from key regions of the southwest Pacific, the data are consistent with a model of trans-Pacific migration, with the islands of Oceania serving as sources rather than sinks for biodiversity.

Caenorhabditis↗

PubMed2002

Circadian rhythm variation in activity, body temperature, and heart rate between C3H/HeJ and C57BL/6J inbred strains.

Inbred mice have been routinely used in studies of genetic effects that determine behavioral variation due to circadian rhythm. In addition to activity patterns (Act), we aimed to characterize variations in the circadian rhythm of deep-body temperature (T(db)) and heart rate (HR) in a specific genetic model of differential cardiorespiratory control. Radiotelemeters were implanted in C3H/HeJ (C3; n = 11) and C57BL/6J (B6; n = 11) inbred strains. Reciprocal first-generation offspring, B6C3F1/J (B6F1; n = 8) and C3B6F1 (C3F1; n = 3) mice, were included to initiate an evaluation of heritable phenotypes. Mice were housed individually in a facility maintained at 23-24 degrees C, and the light-dark cycle was set at 12-h intervals. In each animal, repeated measurements were obtained at 30-min intervals, and the circadian patterns of Act, T(db), and HR were assessed by novel statistical methods that detailed the periodic function for each strain. During the dark phase, B6 mice demonstrated two distinct peaks in Act and T(db) relative to a single early peak for C3 mice. In contrast to the parental strains, B6F1 and C3F1 mice demonstrated intermediate second peaks in Act and T(db). With respect to HR, the C3 strain demonstrated a significantly (P < 0.01) greater daily average compared with B6 mice. The circadian rhythm in HR differed significantly from the Act and T(db) patterns in B6 mice (but not in C3 mice); that is, the periodicity in HR for B6 mice preceded the rise and fall in Act and T(db) during both peaks. The B6 phenotype was also observed in F1 mice. In conclusion, these data suggest that the circadian regulation of Act, T(db), and HR vary significantly among C3, B6, and F1 mice. Furthermore, phenotypic differences between C3 and B6 strains can be used to explore the genetic basis for differential circadian regulation of body temperature and HR.

Activity Cycles↗

PubMed1987

Role of platelets in atherogenesis: relevance to coronary arterial restenosis after angioplasty.

There is now considerable evidence to suggest that some aspects of early lesion formation and later lesion growth are a reaction to injury. Hemodynamic factors are important in determining the site of injury and may produce injury directly. Injury can lead to atherogenesis in animal models as well as in humans. Superficial injury exposes the subendothelium, allowing platelet adhesion, which at high shear rates is dependent on vWF. Platelet adhesion and degranulation release PDGF, which stimulates smooth muscle cell proliferation, synthetic functions, and vasoconstriction. LDL stimulates smooth muscle cell growth as well as damages endothelium in some experimental systems. Thus, a link is provided between platelet and lipid involvement in atherosclerosis. Direct evidence for a role of platelets in atherogenesis comes from studies in which animals were treated to reduce platelet number or function or in which platelet function is genetically impaired (pigs with von Willebrand's disease). In these models, reduced platelet function is associated with less atherosclerosis. Deeper injury exposes collagen, with subsequent platelet aggregation, thrombin and fibrin generation. The role of reduced production of PGI2 and fibrinolytic agents following severe damage is unknown. Deep injury to the vessel occurs during plaque fissuring, the pathologic process underlying most cases of myocardial infarction, unstable angina, and some cases of sudden death. Angioplasty produces amelioration of many patients' symptoms and is safe. However, acute occlusion occurs occasionally, and restenosis in the first year occurs in some 30 percent of patients treated. Angioplasty damages the arterial wall, with endothelial denudation and intimal and medial splitting. Why does this, and plaque injury, by stimulating platelet deposition, not produce more restenosis? Changes in arterial anatomy are likely to be important: the increase in vessel diameter and in blood flow produce conditions less favorable for thrombotic or arteriosclerotic restenosis.

Angioplasty, Balloon↗

PubMed1993

Microcircuitry of posterior cingulate cortex in vitro: electrophysiology and laminar analysis using the current source density method.

We used current source density (CSD) analysis of a laminar profile of subicular stimulus-evoked field potentials recorded in cortical slices in vitro to characterize the interlaminar microcircuitry of posterior cingulate cortex. Neuroanatomic and electrophysiologic data indicate that subiculocingulate tract (SCT) afferents monosynaptically excite apical dendrites of deep laminae (V-VI) neurons, evoking pure EPSPs, while superficial laminae (II/III-IV) neurons are driven polysynaptically, evoking a mixture of longer latency EPSPs and IPSPs. Consistent with this model, CSD analysis of field potential laminar profiles supports the conclusion that activation of excitatory subicular afferent terminal fields in superficial laminae of cingulate cortex elicits primary monosynaptic activation of apical dendrites of deep lamina (V-VI) pyramids. Subsequent EPSP propagation to the somata of these pyramids generated synchronous action potential discharges which appeared to elicit delayed polysynaptic activation of superficial laminae pyramids and interneurons. Latency differences between SCT-stimulus-evoked EPSPs and action potentials in superficial and deep laminae were minimized by stimulus train frequencies of 5-8 Hz, indicating that the proposed microcircuitry can show functional tuning at frequencies characteristic of hippocampal neuronal activity (theta). Such tuning suggests that hippocampal output activity frequency and phase locked to theta rhythm will be preferentially gated through cingulate cortex.

Afferent Pathways↗

PubMed2026

[Applications and Challenges of Deep Learning in Human Genome Research].

In recent years, the advent of high-throughput omics technologies has fueled an explosive growth in human genomic data. Uncovering the latent functions within this vast data has become a significant challenge in functional genomics research. While traditional statistical methods have proved successful for analyzing smaller-scale datasets in the past, they exhibit clear limitations in analytical efficiency and integrating multi-dimensional data, struggling to meet the escalating demands of contemporary genomic analysis. The introduction of deep learning (DL) technologies offers a novel paradigm for this field. This review systematically examines the advances in applying deep learning to human genomics research. Studies demonstrate that when ample labeled data is available, discriminative DL computational methods-such as Convolutional Neural Networks (CNNs) and Long Short-Term Memory networks (LSTMs)-achieve high accuracy and efficiency in genomic variant discovery tasks. Furthermore, generative DL methods, particularly Large Language Models (LLMs) leveraging self-supervised pre-training strategies, effectively integrate complex genomic information and exhibit superior performance in functional genomic sequence annotation and gene regulation studies. This review also explores the application of LLMs in multi-omics data integration and prediction. Looking ahead, the continued accumulation of long-read sequencing and high-dimensional data is expected to enable DL technologies to integrate increasingly complex and heterogeneous genomic information, playing an increasingly crucial role in human genomics research.

Deep Learning↗

PubMed2024

Nanopore sequencing to detect A-to-I editing sites.

Adenosine-to-inosine (A-to-I) RNA editing, mediated by the ADAR family of enzymes, is pervasive in metazoans and functions as an important mechanism to diversify the proteome and control gene expression. Over the years, there have been multiple efforts to comprehensively map the editing landscape in different organisms and in different disease states. As inosine (I) is recognized largely as guanosine (G) by cellular machineries including the reverse transcriptase, editing sites can be detected as A-to-G changes during sequencing of complementary DNA (cDNA). However, such an approach is indirect and can be confounded by genomic single nucleotide polymorphisms (SNPs) and DNA mutations. Moreover, past studies rely primarily on the Illumina platform, which generates short sequencing reads that can be challenging to map. Recently, nanopore direct RNA sequencing has emerged as a powerful technology to address the issues. Here, we describe the use of the technology together with deep learning models that we have developed, named Dinopore (Detection of inosine with nanopore sequencing), to interrogate the A-to-I editome of any organism.

Inosine↗

PubMed1978

Neuronal mechanisms of the late N-wave induced in vitro in thin sections of the olfactory cortex of rats.

Experiments were done to elucidate properties of the late N-wave which was induced in vitro in thin sections of the olfactory cortex of the rat in response to stimulation of the lateral olfactory tract. The late N-wave decreased in size at a stimulation rate of more than once every 90 sec or at temperatures higher than 27 degrees C. The late N-wave was suppressed in the presence of GABA, picrotoxin or bicuculline or in the Cl-free medium. Penicillin or pentylenetetrazol, which blocked actions of GABA on the presynaptic potential, also suppressed the late N-wave. The late N-wave first appeared at postnatal ages of 18--25 days. The late N-wave reversed in polarity when recorded from the deep layers of the sections or from the cut surface of the sections. Single cells in the deep portions of the sections discharged during the late N-wave. Cells in the superficial layers fired just before or after the late N-wave. In order to explain these observations, a neuronal model for generation of the late N-wave was presented.

Action Potentials↗

PubMed2000

Characteristics of plateau activity during the latent period prior to epileptiform discharges in slices from rat piriform cortex.

The deep piriform region has an unusually high seizure susceptibility. Voltage imaging previously located the sites of epileptiform discharge onset in slices of rat piriform cortex and revealed the spatiotemporal pattern of development of two types of electrical activity during the latent period prior to discharge onset. A ramplike depolarization (onset activity) appears at the site of discharge onset. Onset activity is preceded by a sustained low-amplitude depolarization (plateau activity) at another site, which shows little if any overlap with the site of onset. Because synaptic blockade at either of these two sites blocks discharges, it was proposed that both forms of latent period activity are necessary for the generation of epileptiform discharges and that the onset and plateau sites work together in the amplification of electrical activity. The capacity for amplification was examined here by studying subthreshold responses in slices of piriform cortex using two different in vitro models of epilepsy. Under some conditions electrically evoked responses showed a nonlinear dependence on stimulus current, suggesting amplification by strong polysynaptic excitatory responses. The sites of plateau and onset activity were mapped for different in vitro models of epilepsy and different sites of stimulation. These experiments showed that the site of plateau activity expanded into deep layers of neighboring neocortex in parallel with expansions of the onset site into neocortex. These results provide further evidence that interactions between the sites of onset and plateau activity play an important role in the initiation of epileptiform discharges. The site of plateau activity showed little variation with different stimulation sites in the piriform cortex, but when stimulation was applied in the endopiriform nucleus (in the sites of onset of plateau activity), plateau activity had a lower amplitude and became distributed over a much wider area. These results indicate that in the initiation of epileptiform discharges, the location of the circuit that generates plateau activity is not rigidly defined but can exhibit flexibility.

Animals↗

PubMed1998

Comparison of two different delay procedures in a rat skin flap model.

The overall objective of this study was to investigate how a strategic delay procedure could generate large flaps survival. On the basis of the vascular anatomy in 10 rats, a long three-territory skin flap spanning the length of the rat dorsum was designed. This flap was distally based on the deep circumflex iliac artery. The adjacent territories consisted of a large perforator of the posterior intercostal artery and the lateral thoracic artery in sequence. Two different vascular delay procedures were used and compared in nine animals. One dorsal midline incision was used to perform the two different delay procedures. The limited delay technique was performed by ligating a large cutaneous perforator of the posterior intercostal artery on one side of the rat. The extensive delay procedure was completed by ligating the cutaneous perforators of the posterior intercostal artery and the lateral thoracic artery on the other side of the rat. After a delay period of 10 days, the paired flaps were elevated, respectively, and sutured back in place. Seven days later, the area of viable skin flap was measured by the paper template technique. The animals were then killed, and the dorsal skin arteriograms were obtained by injecting a lead oxide mixture. Vascular changes of the paired flaps were assessed. In the extensive delay group, 100-percent survival was seen in four animals and distal partial necrosis was observed in five animals. The average survival area in the extensive delay group was 85.5 +/- 14.2 percent (mean +/- SD), whereas the flaps in limited delay group showed 100-percent survival in all animals (p < 0.05). In flaps that survived completely, the choke vessels among three vascular territories anastomosed throughout the flap. In the partially necrotic flaps, some choke and true anastomoses existed between the deep circumflex iliac and the posterior intercostal artery territories and the flap necrosis occurred in the third territory. This study suggests that the limited delay technique (ligation of one artery in the territory adjacent to the base of the flap) was the more effective delay procedure in the rat dorsal skin flap model.

Animals↗

PubMed1999

Host-induced, stage-specific virulence gene activation in Candida albicans during infection.

An understanding of the complex interactions between pathogenic microbes and their host must include the identification of gene expression patterns during infection. To detect the activation of virulence genes in the opportunistic fungal pathogen Candida albicans in vivo by host signals, we devised a reporter system that is based on FLP-mediated genetic recombination. The FLP gene, encoding the site-specific recombinase FLP, was genetically modified for expression in C. albicans and fused to the promoter of the SAP2 gene that codes for one of the secreted aspartic proteinases, which are putative virulence factors of C. albicans. The SAP2P-FLP fusion was integrated into one of the SAP2 alleles in a strain that contained a deletable marker that conferred resistance to mycophenolic acid and was flanked by direct repeats of the FLP recognition target (FRT). Using this reporter system, a transient gene induction could be monitored at the level of single cells by the mycophenolic acid-sensitive phenotype of the colonies generated from such cells after FLP-mediated marker excision. In two mouse models of disseminated candidiasis, SAP2 expression was not observed in the initial phase of infection, but the SAP2 gene was strongly induced after dissemination into deep organs. In contrast, in a mouse model of oesophageal candidiasis in which dissemination into internal organs did not occur, no SAP2 expression was detected at any time. Our results support a role of the SAP2 gene in the late stages of an infection, after fungal spread into deep tissue. This new in vivo expression technology (IVET) for a human fungal pathogen allows the detection of virulence gene induction at different stages of an infection, and therefore provides clues about the role of these genes in the disease process.

Animals↗

PubMed1998

Morphometry of human neck muscles.

STUDY DESIGN: Cadaveric dissections were used to study muscle morphometry. OBJECTIVE: To describe systematically the musculotendinous lengths, fascicle lengths, pennation angles, and physiologic cross-sectional areas of neck and shoulder muscles implicated in head movement. SUMMARY OF BACKGROUND DATA: In previous studies of neck-muscle anatomy, researchers described only a subset of muscle features, often using crude or indirect methods. None used microdissected muscles to correct measured parameters for the presence of multiple fiber compartments, internal aponeuroses, or variations in fiber or sarcomere length required for qualitative models of force-generating capabilities. METHODS: Muscle mass, pennation angle, fascicle length, and sarcomere length were measured in 14 neck muscles from 10 human cadavers. Architecturally complex muscles with multiple attachments were divided into subvolumes, and each subvolume was examined from both the superficial and deep surfaces, Internal aponeuroses were microdissected within muscles to characterize architectural specializations. Physiologic cross-sectional areas were calculated from the morphometric data. RESULTS: The neck musculature was architecturally complex. Many muscles crossed two or more joints and had multiple attachments to different bones. In some, the presence of tendons and aponeuroses was associated with specializations in fascicle organization. Considerable interindividual variation was found in the number and location of tendinous insertions of the scalenes and longissimus capitis muscles. In addition, rhomboideus showed significant variations in its size and shape. The cross-sectional areas of neck muscles from large and small subjects did not scale proportionately with body height and weight, nor did individual muscles with widely varying cross-sectional areas (0.3-15.3cm2) scale from on subject to another. CONCLUSIONS: The accuracy of morphometry can be improved by incorporating measurements made by microdissecting neck muscles. The presence of aponeurotic attachments can greatly shorten fascicle length; failure to identify such attachments can lead to underestimates of cross-sectional areas. Accuracy of a generalized model of the neck is also improved by normalizing sarcomere lengths in all muscles.

Aged↗

PubMed2001

[Anatomic principles of urinary incontinence].

The morphological fundamentals of urinary continence are still subject to controversy. This was the reason for a renewed examination of the sphincter musculature of the lower urinary tract. This study included 50 male and 15 female autopsy specimens. The organs of the lower urinary tract including the neighboring organs had been removed in their entirety and histologically reprocessed en bloc as a complete series of sections. We were able to demonstrate that the internal sphincter or m. sphincter vesicae is represented as a circular, distinct structure which elliptically embraces the internal urethral orifice. Lamellas of the detrusor are not involved in the formation of the internal sphincter. In females and males, the external sphincter consists of a striated and a smooth muscular part (m. sphincter urethrae transversostriatus et glaber). In transverse sections, the muscle has a horseshoe shape. It is completely separated by connective tissue from the musculature of the pelvic floor. A deep transverse perineal muscle does not exist. The histological findings were used for the construction of a digital three-dimensional model of the anatomy of the lower urinary tract. Computer animations of the model with integrated original histologies were generated and stored as a computer video on a CD-ROM attached to this journal.

Computer Simulation↗

PubMed1995

Susceptibility of different cell layers of the anterior and posterior part of the piriform cortex to electrical stimulation and kindling: comparison with the basolateral amygdala and "area tempestas".

Several lines of evidence suggest that the piriform cortex functions as a generator in the development and propagation of forebrain (limbic type) seizures, particularly in the kindling model of epilepsy. It is, however, not clear where, within the rather large piriform cortex region, the generator resides, and how much tissue is involved. Highly sensitive loci to chemical or electrical stimulation have been described both in the deep anterior and posterior parts of the piriform cortex. Furthermore, data from piriform cortex slice preparations indicated that epileptiform potentials originate in deep structures, particularly the endopiriform nucleus that underlies the piriform cortex. In the present study, in rats, we implanted stimulation and recording electrodes in various rostrocaudal locations of the piriform cortex and endopiriform nucleus, including the "area tempestas", i.e. a structure in the anterior part of the piriform cortex previously proposed to be critically involved in the generation of convulsive seizures of limbic origin. Within the piriform cortex, electrodes were aimed at different cellular layers of this structure. For comparison, additional animals received electrodes in different parts of the basolateral amygdala. A total of 19 different locations was obtained in this way. The susceptibility of these locations to electrical stimulation was characterized by determining the threshold for induction of afterdischarges. The afterdischarge threshold was lowest in layer III of the posterior piriform cortex and some locations in the endopiriform nucleus, whereas amygdala and "area tempestas" displayed higher values. In several animals, particularly those with electrodes in layer III of the posterior piriform cortex, spontaneous spiking was seen in prestimulation recordings, whereas this was never observed in recordings from the amygdala. Subsequent kindling by repeated stimulation of the various locations demonstrated marked differences in afterdischarge threshold reduction and kindling rate. The most marked decreases in afterdischarge threshold were seen in locations within layer III of the piriform cortex, whereas several other locations, including the "area tempestas", exhibited only moderate decreases or no decrease at all. In contrast to previous observations with only few locations in the piriform cortex region, the posterior piriform cortex was not in general slower to kindle than the anterior piriform cortex, although some locations in the posterior piriform cortex exhibited significantly lower kindling rates than the amygdala. The highest kindling rate was seen in the dorsal endopiriform nucleus.(ABSTRACT TRUNCATED AT 400 WORDS)

Amygdala↗

PubMed2001

Prediction of electroencephalographic spectra from neurophysiology.

A recent neurophysical model of propagation of electrical waves in the cortex is extended to include a physiologically motivated subcortical feedback loop via the thalamus. The electroencephalographic spectrum when the system is driven by white noise is then calculated analytically in terms of physiological parameters, including the effects of filtering of signals by the cerebrospinal fluid, skull, and scalp. The spectral power at low frequencies is found to vary as f(-1) when awake and f(-3) when asleep, with a breakpoint to a steeper power-law tail at frequencies above about 20 Hz in both cases; the f(-1) range concurs with recent magnetoencephalographic observations of such a regime. Parameter sensitivities are explored, enabling a model with fewer free parameters to be proposed, and showing that spectra predicted for physiologically reasonable parameter values strongly resemble those observed in the laboratory. Alpha and beta peaks seen near 10 Hz and twice that frequency, respectively, in the relaxed wakeful state are generated via subcortical feedback in this model, thereby leading to predictions of their frequencies in terms of physiological parameters, and of correlations in their occurrence. Subcortical feedback is also predicted to be responsible for production of anticorrelated peaks in deep sleep states that correspond to the occurrence of theta rhythm at around half the alpha frequency and sleep spindles at 3/2 times the alpha frequency. An additional positively correlated waking peak near three times the alpha frequency is also predicted and tentatively observed, as are two new types of sleep spindle near 5/2 and 7/2 times the alpha frequency, and anticorrelated with alpha. These results provide a theoretical basis for the conventional division of EEG spectra into frequency bands, but imply that the exact bounds of these bands depend on the individual. Three types of potential instability are found: one at zero frequency, another in the theta band at around half the alpha frequency, and a third at the alpha frequency itself.

Adult↗

PubMed1997

Increased excitability and inward rectification in layer V cortical pyramidal neurons in the epileptic mutant mouse Stargazer.

The excitability of layer V cortical pyramidal neurons was studied in vitro in the single-locus mutant mouse stargazer (stg), a genetic model of spike wave epilepsy. Field recordings in neocortical slices from mutant mice bathed in artificial cerebrospinal fluid revealed spontaneous synchronous network discharges that were never present in wild-type slices. Intracellular and whole cell recordings from stg/stg neurons in deep layers showed spontaneous giant depolarizing excitatory post-synaptic potentials generating bursts of action potentials, and a 78% reduction in the afterburst hyperpolarization. Whole cell recordings revealed gene-linked differences in active membrane properties in two types of regular spiking neurons. Single action potential rise and decay times were reduced, and the rheobase current was decreased by 68% in mutant cells. Plots of spike frequency-current relationships revealed that the gain of this relation was augmented by 29% in the mutant. Comparisons of visually identified pyramidal neurons firing properties in both genotypes revealed no difference in single action potential afterhyperpolarization. Voltage-clamp recordings showed an approximately threefold amplitude increase in a cesium-sensitive inward rectifier. No cell density or soma size differences were observed in the layer V pyramidal neuron population between the two genotypes. These results demonstrate an autonomous increase in cortical network excitability in a genetic epilepsy model. This defect could lower the threshold for aberrant thalamocortical spike wave oscillations in vivo, and may contribute to the mechanism of one form of inherited absence epilepsy.

Action Potentials↗

PubMed2023

Active learning of enhancer and silencer regulatory grammar in photoreceptors.

Cis-regulatory elements (CREs) direct gene expression in health and disease, and models that can accurately predict their activities from DNA sequences are crucial for biomedicine. Deep learning represents one emerging strategy to model the regulatory grammar that relates CRE sequence to function. However, these models require training data on a scale that exceeds the number of CREs in the genome. We address this problem using active machine learning to iteratively train models on multiple rounds of synthetic DNA sequences assayed in live mammalian retinas. During each round of training the model actively selects sequence perturbations to assay, thereby efficiently generating informative training data. We iteratively trained a model that predicts the activities of sequences containing binding motifs for the photoreceptor transcription factor Cone-rod homeobox (CRX) using an order of magnitude less training data than current approaches. The model's internal confidence estimates of its predictions are reliable guides for designing sequences with high activity. The model correctly identified critical sequence differences between active and inactive sequences with nearly identical transcription factor binding sites, and revealed order and spacing preferences for combinations of motifs. Our results establish active learning as an effective method to train accurate deep learning models of cis-regulatory function after exhausting naturally occurring training examples in the genome.

Journal Article↗