Biology subjects

Ball, S. T. M.

Publications and source records attributed to Ball, S. T. M..

5 recordsLinked to original sources

bioRxiv2026

DiffuChannel: A Conditional Denoising Diffusion Model that Synthesises Single-Molecule Patch-Clamp Data with Exactly Known Idealisation (Ground Truth)

Development of computational models for single-channel patch-clamp analysis is hampered by limited availability of labelled (idealised) data. Since experiment alone cannot provide a genuine ground truth, two simulation methods are commonly employed; Stochastic sampling from a specified Markov model, augmented with synthetic noise allows creation of unlimited, flawlessly idealised/annotated datasets, yet its authenticity hinges entirely on the premises underlying data and noise generation. A more recent method uses GAN-based generation, this also replicates real-world noise and artefacts, but does not include a ground truth idealisation or annotation and training is notoriously unstable. Here, we introduce DiffuChannel, a conditional noise-removal diffusion model that addresses these drawbacks. Channel dynamics come directly from a user-defined discrete-time Markov chain (or a model fitted to biological seed data), thus, the state assignment for each synthetic trace is precisely determined by design. The diffusion model is then trained exclusively to capture the noise distribution of raw current conditioned on a given idealisation, thereby replicating genuine noise, filtering, and artefact patterns observed in real data without needing to hand craft noise models. We use a two-step coarse-to-fine workflow that starts with a defined slow-envelope component conditioned on longer-term dwell-time features extracted from the input idealisation. This allows a degree of baseline drift and time-dependent relaxations to be captured over timescales far longer than the network's field of view. At generation time, we use a context-preserving inpainting method that enables creation of relatively seamless traces of any desired duration. We demonstrate that with only a limited set of labelled seed traces, synthetic records yield realistic all-points amplitude distribution, noise power spectrum, autocorrelation pattern, and slow envelope comparable to the source data. The model achieves a maximum mean discrepancy (MMD) between windowed real and generated data of 0.005 (median across the five phenotypes). This compares to a within real-split-half MMD value of 0.002 and an amplitude-matched white noise MMD value of 0.108. In this paper, we benchmark this against DeepGANnel. Since the underlying Markov model is defined rather than estimated, DiffuChannel generates infinite, perfectly labelled, phenotype-consistent training data ideal for developing and testing idealisation algorithms. We find that, in practice, this new method is far more stable than previous GAN models and runs start to end without human intervention. We provide the model open-source along with an interactive web-based training and synthesis interface/app.

biophysics↗

bioRxiv2024

Domain Specific AI Segmentation of IMPDH2 Rod/Ring Structures in Mouse Embryonic Stem Cells

BackgroundInosine monophosphate dehydrogenase 2 (IMPDH2) is an enzyme that catalyses the rate limiting step of guanine nucleotides. In mouse embryonic stem cells (ESCs) IMPDH2 is held as large multi-protein complexes known as rod-ring (RR) structures that dissociate when ESCs differentiate. Manual analysis of RR structures from confocal microscopy images, although possible, is not feasible on a large scale due to the quantity of RR structures present in each field of view. To address this analysis bottleneck, we have created a fully automatic RR image classification pipeline to segment, characterise and measure feature distributions of these structures in ESCs. ResultsWe find that this model can automatically segment images with a Dice score of over 80% for both rods and rings for in-domain images compared to expert annotation, with a slight drop to 70% for datasets out of domain. Important feature measurements derived from these segmentations show high agreement with the measurements derived from expert annotation, achieving an R2 score of over 90% for counting the number of rings and rods over the dataset. ConclusionsWe have established for the first time a quantitative baseline for RR distribution in pluripotent ESCs and have made a pipeline available for training to be applied to other models in which RR remain an open topic of study.

bioinformatics↗

bioRxiv2024

Peroxiredoxin 2 mediates redox-stimulated adaptations to oxidative phosphorylation induced by contractile activity in human skeletal muscle myotubes

Skeletal muscle generates superoxide during contractions, which is converted to hydrogen peroxide (H2O2). H2O2 has been proposed to activate signalling pathways and transcription factors that regulate adaptive responses to exercise, but the concentration required to oxidize and activate key redox-sensitive signalling proteins in vitro is much higher than the typical intracellular levels seen in muscle after exercise. We hypothesized that 2-Cys-peroxiredoxins (PRDX), which rapidly oxidize in the presence of physiological concentrations of H2O2, serve as intermediary signalling molecules and play a crucial role in activating adaptive pathways following muscle contractions. This study has examined the human muscle myotube responses to contractile activity, or exposure to low extracellular concentrations (2.5-5 {micro}M) of H2O2 and whether knock down of muscle PRDX2 alters the differential gene expression (DEG) that results from these stresses. Exposure of human skeletal muscle myotubes to a 15 min period of aerobic electrically stimulated isometric contractions or 5M H2O2 induced substantial changes in DEG with modification of many genes associated with adaptations of skeletal muscle to contractile activity. Common DEG in these conditions included upregulation of genes associated with increased mitochondrial oxidative phosphorylation, including COX1, COX2, COX3 and ATP6. In myotubes with PRDX2 knock down (94% decrease in PRDX2 mRNA), the upregulation of genes associated with increased mitochondrial oxidative phosphorylation was abolished following contractile activity or exposure to H2O2. These data indicate that a common effect of contractile activity and exposure to "physiological" levels of H2O2 in human myotubes is to increase the expression of multiple genes associated with increased mitochondrial oxidative phosphorylation. Furthermore, these effects were abolished in PRDX2 knock down myotubes indicating that adaptations to upregulate multiple genes related to increased mitochondrial capacity in human muscle myotubes in response to exercise is both redox regulated and requires PRDX2 as an essential mediator of the effects of H2O2.

physiology↗

bioRxiv2020

DeepGANnel: Synthesis of fully annotated single molecule patch-clamp data using generative adversarial networks

Development of automated analysis tools for "single ion channel" recording is hampered by the lack of available training data. For machine learning based tools, very large training sets are necessary with sample-by-sample point labelled data (e.g., 1 sample point every 100microsecond). In an experimental context, such data are labelled with human supervision, and whilst this is feasible for simple experimental analysis, it is infeasible to generate the enormous datasets that would be necessary for a big data approach using hand crafting. In this work we aimed to develop methods to generate simulated ion channel data that is free from assumptions and prior knowledge of noise and underlying hidden Markov models. We successfully leverage generative adversarial networks (GANs) to build an end-to-end pipeline for generating an unlimited amount of labelled training data from a small annotated ion channel "seed" record, and this needs no prior knowledge of theoretical dynamical ion channel properties. Our method utilises 2D CNNs to maintain the synchronised temporal relationship between the raw and idealised record. We demonstrate the applicability of the method with 5 different data sources and show authenticity with t-SNE and UMAP projection comparisons between real and synthetic data. The model would be easily extendable to other time series data requiring parallel labelling, such as labelled ECG signals or raw nanopore sequencing data.

physiology↗

bioRxiv2020

Discriminant Analysis of Principle Component analyses of Physiological Data.

There are many situations in physiological and pharmacological analyses where multivariate data is collected. Frequently these are analysed with t-tests and multiple (Bonferroni) comparisons or ANOVA with post-hoc test. Increasingly, even with more powerful computers many variables and it seems that feature reduction would be a useful approach. The most commonly used method is principle component analyses, but in this report we compare this to a technique developed for genetic analyses, discriminant analysis of principle component (DAPC) analyses. A simple to use and well-maintained library exists for DAPC analyses, Adegenet2, and using this we find that DAPC detects differences between synthetic physiological datasets with significantly greater accuracy than traditional PCA.

physiology↗