Biology subjects

Kunz, P.

Publications and source records attributed to Kunz, P..

3 recordsLinked to original sources

bioRxiv2024

Fast FEM-based Electric Field Calculations for Transcranial Magnetic Stimulation

ObjectiveTo provide a Finite-Element Method (FEM) for rapid, repeated evaluations of the electric field induced by transcranial magnetic stimulation (TMS) in the brain for changing coil positions. ApproachPreviously, we introduced a first-order tetrahedral FEM enhanced by super- convergent patch recovery (SPR), striking a good balance between computational efficiency and accuracy (Saturnino et al 2019 J. Neural Eng. 16 066032). In this study, we refined the method to accommodate repeated simulations with varying TMS coil position. Starting from a fast direct solver, streamlining the pre- and SPR-based post-calculation steps using weight matrices computed during initialization strongly improved the computational efficiency. Additional speedups were achieved through efficient multi-core and GPU acceleration, alongside the optimization of the volume conductor model of the head for TMS. Main ResultsFor an anatomically detailed head model with [~]4.4 million tetrahedra, the optimized implementation achieves update rates above 1 Hz for electric field calculations in bilateral gray matter, resulting in a 60-fold speedup over the previous method with identical accuracy. An optimized model without neck and with adaptive spatial resolution scaled in dependence to the distance to brain grey matter, resulting in [~]1.9 million tetrahedra, increased update rates up to 10 Hz, with [~]3% numerical error and [~]4% deviation from the standard model. Region-of-interest (ROI) optimized models focused on the left motor, premotor and dorsolateral prefrontal cortices reached update rates over 20 Hz, maintaining a difference of <4% from standard results. Our approach allows efficient switching between coil types and ROI during runtime which greatly enhances the flexibility. SignificanceThe optimized FEM enhances speed, accuracy and flexibility and benefits various applications. This includes the planning and optimization of coil positions, pre-calculation and training procedures for real-time electric field simulations based on surrogate models as well as targeting and dose control during neuronavigated TMS.

neuroscience↗

bioRxiv2024

Prediction of protein biophysical traits from limited data: a case study on nanobody thermostability through NanoMelt

1In-silico prediction of protein biophysical traits is often hindered by the limited availability of experimental data and their heterogeneity. Training on limited data can lead to overfitting and poor generalisability to sequences distant from those in the training set. Additionally, inadequate use of scarce and disparate data can introduce biases during evaluation, leading to unreliable model performances being reported. Here, we present a comprehensive study exploring various approaches for protein fitness prediction from limited data, leveraging pre-trained embeddings, repeated stratified nested cross-validation, and ensemble learning to ensure an unbiased assessment of the performances. We applied our framework to introduce NanoMelt, a predictor of nanobody thermostability trained with a dataset of 640 measurements of apparent melting temperature, obtained by integrating data from the literature with 129 new measurements from this study. We find that an ensemble model stacking multiple regression using diverse sequence embeddings achieves state-of-the-art accuracy in predicting nanobody thermostability. We further demonstrate NanoMelts potential to streamline nanobody development by guiding the selection of highly stable nanobodies. We make the curated dataset of nanobody thermostability freely available and NanoMelt accessible as a downloadable software and webserver. 2 Significance StatementRapidly predicting protein biophysical traits with accuracy is a key goal in protein engineering, yet efforts to develop reliable predictors are often hindered by limited and disparate experimental measurements. We introduce a framework to predict biophysical traits using few training data, leveraging diverse machine learning approaches via a semi-supervised framework combined with ensemble learning. We applied this framework to develop NanoMelt, a tool to predict nanobody thermostability trained on a new dataset of apparent melting temperatures. Nanobodies are increasingly important in research and therapeutics due to their ease of production and small size, which allows deeper tissue penetration and seamless combination into multi-specific compounds. NanoMelt outperforms available methods for protein thermostability prediction and can streamline nanobody development by guiding the design and selection of highly stable nanobodies during discovery and optimization campaigns.

bioengineering↗

bioRxiv2018

transcranial Direct Current Stimulation Studies Open Database (tDCS-OD)

Transcranial direct current stimulation (tDCS) is increasingly evaluated for a myriad of clinical indications and performance enhancement mainly related to cognitive and motor applications. As a result tDCS has been tested with diverse inclusion criteria and subject populations. tDCS is customized for applications by many factors, including stimulation dose (montage, current, time), associated training and subject inclusion/exclusion criteria. The rapid proliferation of applications, technological advancements and emerging scientific discoveries presents challenges to the organization and consolidation of tDCS data - which can lead to scientific, clinical and public confusions. In this paper, we develop a system to summarize and consolidate methodological aspects of tDCS, concentrating on study design and the physical parameters of the stimulation. We introduce a community-driven, open access database, where these parameters, including stimulation intensities, duration, electrode sizes, montages, and subject information, are noted for relevant tDCS publications. The transcranial Direct Current Stimulation Studies Open Database (tDCS-OD) (http://tdcsdatabase.com) will support constructive dialogue on research and clinical tDCS applications, critical evaluation of past work, and identification of promising protocols, while reducing ambiguity about stimulation methodology. The database design allows ongoing updates and editing, with transparent version control. At the moment, the database includes information about 56,613 tDCS sessions.

neuroscience↗