Biology subjects

Linkevicius, D.

Publications and source records attributed to Linkevicius, D..

2 recordsLinked to original sources

bioRxiv2025

One model to rule them all: unification of voltage-gated potassium channel models via deep non-linear mixed effects modelling

Ion channels are essential for signal processing and propagation in neural cells. Voltage-gated ion channels permeable to potassium (Kv) form one of the most prominent channel families. Techniques used to model the voltage-dependent gating of Kv channels date back to Hodgkin and Huxley (1952). Different Kv types can display radically different kinetic properties, requiring different mathematical models. However, the construction of Hodgkin-Huxley-like (HH-like) models is generally complex and time consuming due to the number of parameters, their tuning and having to choose functional forms to model gating. In addition to the between-Kv type heterogeneity, there can be significant within-Kv type kinetic heterogeneity between different cells with genetically identical channels. Since HH-like models do not account for such variability, extensions to it are necessary. We use scientific machine learning (SciML), the integration of machine learning methodologies with existing scientific models, and non-linear mixed effects (NLME) modelling to bypass the limitations of HH-like modelling. NLME is a modelling methodology that takes into account both within- and between-subject variability. These tools allowed us to complement the HH-like modelling and construct a unified SciML HH-like model that fits the recordings from 20 different Kv types. The unified SciML HH-like model produced closer fits to the data compared to a set of seven previous HH-like models and was able to represent the highly heterogeneous data from different cells. Our model may be the first step in producing a SciML foundation model for ion channels that would be capable of modelling the gating kinetics of any ion channel type. Author summaryIon channels are complex molecules embedded in the membranes of neurons - the cells responsible for signal propagation and processing in the brain. Ion channels can open and close in response to various types of stimuli, in particular the voltage difference across the cell membrane. Computational modelling, usage of mathematical techniques to represent a system and algorithmically solve for its dynamics, has been previously used to understand the dynamics of voltage-gated ion channels. However, computational modelling of voltage-gated ion channels requires costly and complex optimization routines to optimize their structure and parameters. We utilize two tools new to the modelling of voltage-gated ion channels - scientific machine learning and non-linear mixed effects modelling - to bypass some limitations associated with the existing methods. By using scientific machine learning and non-linear mixed effects modelling we were able to create a unified model capable of modelling the gating dynamics of 20 different ion channels. This is in stark contrast to the existing modelling approaches, where each channel requires its own model. Moreover, our unified model performed better than seven existing ion channel gating models. Therefore, the tools we used and the model we created is a significant step forward in facilitating the modelling of ion channel gating. Future work could include even more ion channels types within the scope of our unified model.

neuroscience↗

bioRxiv2024

Fitting and comparison of calcium-calmodulin kinetic schemes to a common data set using non-linear mixed effects modelling

Calmodulin is a calcium binding protein that is essential in calcium signalling in the brain. There are many computational models of calcium-calmodulin binding that capture various calmodulin features. However, existing models have generally been fit to different data sets, with some publications not reporting their training and validation performance. Moreover, there is no model comparison using a common benchmark data set as is common practice in other modeling domains. Finally, some calmodulin models have been fit as a part of a larger kinetic scheme, which may have resulted in parameters being underdetermined. We address these three limitations of previous models by fitting the published calcium-calmodulin schemes to a common calcium-calmodulin data set comprising of equlibrium data from Shifman et al. and dynamical data from Faas et al. Due to technical limitations, the amount of uncaged calcium in Faas et al. data could not be predicted with certainty. To find good parameter fits, despite this uncertainty, we used non-linear mixed effects modelling as implemented in the Pumas.jl package. The Akaike information criterion values for our reaction rates were significantly lower than for the published parameters, indicating that the published parameters are suboptimal. Moreover, there were significant differences in calmodulin activation, both between the schemes and between our reaction rate and those previously published. A kinetic scheme with independent lobes and unique, rather than identical, binding sites fit the data best. Our results support two hypotheses: (1) partially bound calmodulin is important in cellular signalling; (2) calcium binding sites within a calmodulin lobe are kinetically distinct rather than identical. We conclude that more attention should be given to validation and comparison of models of individual molecules. Author summaryLearning and memory depend on changes in synapses, the connections between nerve cells. To understand how learning and memory work, it is important to understand how the proteins involved in these changes are regulated. Computational modelling of biochemical reactions has been used to understand the regulation processes involved in learning and memory. However, computational models often rely on simplifications of biochemical reaction schemes, and it can be difficult to tell which simplifications are "the best", i.e. capture important aspects of a proteins behaviour. It is also difficult to estimate the best model parameters, such as binding reaction rates or binding strengths. Modellers often base their estimates on known experimental results, but often not in a structured way. We examine and compare a variety of computational models of calmodulin, a protein necessary for changes in synaptic strength. We apply a new approach to infer the model parameters that best fit two published datasets, and compare our parameters to previously published ones. We find that some simplified reaction schemes are more successful than others in capturing the activation pattern of calmodulin. We also find that the parameter sets found through our approach outperform previously published parameter sets in fitting the experimental data.

systems biology↗