Daily Summary for 2026-10-21

Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.

Ines: It's the twenty-first of October, twenty twenty-six, and this is the day's research.

Marcus: 15 new papers came out today.

Ines: Hi, I'm Ines.

Marcus: And I'm Marcus, here with Yuki.

Ines: We'll take the day in one pass, then pull out the papers we're staying with.

Marcus: Start us off with Kinetics of template-directed multistate copolymerization.

The summary: Ines: Welcome to the twenty first of October, twenty twenty six. Today we focus on better understanding calcium signaling dynamics in cell culture using data driven modeling because clearer pictures allow us to observe systems with greater detail.

Marcus: We are comparing two methods for system identification and parameter discovery for governing ordinary differential equations.

Yuki: Specifically the Constrained Regularized Least-Squares Method CRLSM and Physics Informed Neural Networks PINN.

Ines: This comparison is in the work titled Data driven multi agent modelling of calcium interactions in cell culture PINN vs Regularized Least-squares.

Marcus: What does this comparison mean for our understanding?

Yuki: It helps us characterize biological systems with greater detail by providing clearer pictures of these processes.

Ines: So the focus is on how we are trying to better understand calcium signaling dynamics in cell culture using data driven modeling.

Marcus: The methods being compared are CRLSM and PINN for discovering parameters in ordinary differential equations.

Yuki: The study uses these methods to see if one provides a clearer picture of the underlying system behavior.

Ines: We are comparing those two approaches in the paper Data driven multi agent modelling of calcium interactions in cell culture PINN vs Regularized Least-squares.

Marcus: Does this comparison illuminate how we observe and characterize biological systems with greater detail?

Yuki: Yes getting a clearer picture of these processes allows us to see biological systems with greater detail.

Ines: How does digital monitoring inform clinical practice?

Marcus: Establishing the Minimal Clinically Important Difference for Smartphone-Derived Gait Measures in Multiple Sclerosis sets concrete thresholds.

Yuki: This relates to data-driven discovery, comparing physics-informed neural networks and regularized least squares methods.

Ines: That comparison identifies parameters governing calcium interactions in cell culture.

Marcus: The most significant finding concerns how receptive field shapes in the primary visual cortex might vary.

Yuki: Do the receptive fields span variability over the degree of elongation?

Ines: This suggests biological receptive field shapes should span degrees of freedom from affine image transformations.

Marcus: It supports affine covariance across a population.

Yuki: This builds on Tony Lindeberg's proposal for simple cells under various transformations.

Ines: His work in AKMPC suggested receptive fields should be covariant under uniform spatial scaling and temporal scaling.

Marcus: This covariance concept is examined in orientation selectivity properties for integrated affine quasi quadrature models of complex cells.

Yuki: That shows assuming spatial receptive fields are covariant implies they must span variability over the degree of elongation.

Ines: PINNs failed to match CRLSM performance today.

Marcus: They did not provide fair parameter estimation under their setup.

Yuki: This suggests room for improvement through tuning and uncertainty quantification.

Ines: PINNs are promising but need more careful tuning for future research.

Marcus: Agreed, uncertainty quantification is key now to see real gains.

Yuki: Data-driven multi-agent modelling of calcium interactions in cell culture: PINN vs Regularized Least-squares is worth hearing because it compares these methods for modeling calcium delivery.

Ines: That comparison helps us understand the trade-offs clearly.

Marcus: It allows better observation and characterization of processes like calcium signaling in cell culture.

Yuki: Data-driven discovery of dynamics in biological systems is valuable for observing such complex processes.

Lucky paper: 2606.22448: Ines: We're looking at Kinetics of template-directed multistate copolymerization today because it shows how we can get exact solutions for growth velocity and sequence probabilities.

Marcus: This paper uses a Markov jump process where the system states include the catalyst, template, and its copy.

Yuki: The mathematical formulation describes these as linear ordinary differential equations governing the time evolution of probabilities for finding the system in any one of its particular states.

Ines: So they are using a matrix factorization ansatz pioneered by Coleman and Fox to solve these kinetic equations.

Marcus: This leads to an iterated matrix function system running backward along the template, which is then combined with a forward iteration.

Yuki: That combination gives them the exact solution for the mean growth velocity and the sequence probabilities of the copy grown along a given template.

Ines: The paper notes different regimes depending on whether that template is periodic or disordered.

Marcus: If it's periodic, there are two generic regimes: a stationary regime where mean length converges to a finite value, and a steady growth regime where mean length increases linearly with time.

Yuki: But if the template is disordered, an intermediate regime exists where the mean length manifests sublinear growth in time like hlit ∼ t gamma with an exponent between zero and one.

Ines: The numerical validation compares this iterative matrix function system to a simulation using Gillespie’s algorithm for the corresponding stochastic process.

Marcus: The IMFS method is more than a million times faster than the numerical simulation method based on Gillespie’s algorithm, while always giving the exact asymptotic solution of the problem.

Yuki: This comparison confirms that both methods provide exact values for the mean growth velocity and local probabilities of monomeric units for the copy grown along a given template.

Ines: The conclusion is that this IMFS method provides an exactly asymptotic solution of the kinetic equations ruling the stochastic process.

Marcus: They show that in steady growth, if attachment rates are too small with respect to detachment rates, the growth of the copy is stalled.

Yuki: The paper also explains how to calculate the stationary probability distribution of the length of the copy using those backward and forward iterations.

Ines: This exact solution for mean growth velocity is a key result from Kinetics of template-directed multistate copolymerization.

Marcus: It really shows how powerful this iterative method is in providing an exact asymptotic solution to the kinetic equations.

Yuki: And it also shows that when attachment rates are too small compared to detachment rates, the copy growth stalls.

Lucky paper: 2601.13714: Ines: We're moving into a piece on cost-effectiveness and vaccination strategies in Korea, specifically the paper Cost-Effectiveness of Adult Hepatitis A Vaccination Strategies in Korea Under an Aging Susceptibility Profile.

Marcus: This study looks at how to prioritize adult hepatitis A vaccination when we look at resource constraints.

Yuki: The core problem is that hepatitis A severity increases sharply with age, and Korea is seeing a shift where low-seroprevalence adult cohorts are aging into higher fatality risk groups.

Ines: So the paper is addressing this demographic and immunological transition through an age-structured transmission model calibrated to Korean data.

Marcus: What kind of model did they use for this simulation-based cost-effectiveness analysis?

Yuki: They stratified the population into eight ten-year age groups and incorporated Korea’s National Immunization Program to reflect real vaccination dynamics.

Ines: They tested three adult vaccination scenarios: S1 covering twenty to thirty-nine years, S2 focusing on forty to fifty-nine years with pre-vaccination antibody testing, and S3 covering the full range of twenty to fifty-nine years with antibody testing applied specifically to those aged over forty.

Marcus: The optimal control theory they used was designed to derive monthly vaccination allocation trajectories for the period from two thousand twenty-five through two thousand fifty-four.

Yuki: They minimized an objective functional that included disease burden and socioeconomic costs, computing disability-adjusted life years DALYs over an evaluation horizon of eighty-three years discounted at a continuous annual rate of four point five percent.

Ines: The key findings show that targeting adults aged forty to fifty-nine years, scenario S2, yielded the most favorable and robust cost-effectiveness profile across both healthcare system and societal perspectives.

Marcus: That S2 strategy resulted in an incremental cost effectiveness ratio ICER of USD fifty-two thousand one hundred sixty-six per DALY averted within the healthcare system perspective.

Yuki: It achieved a sizable reduction in DALYs with seven thousand four hundred fifty-eight DALYs averted while incurring only a moderate increase in costs.

Ines: Scenario S3 covered the largest absolute reduction in DALYs but required substantially higher incremental costs when compared to S2.

Marcus: The optimal control strategies showed how doses would be preferentially allocated across age groups under each strategy, and specifically in scenario S3, doses were distributed almost evenly across adults from 20s to 50s from two thousand twenty-five through the early 2040s.

Yuki: The societal perspective results showed higher incremental costs for both S2 and S3 because it included extra components like productivity losses and non-medical expenses.

Ines: The paper concludes that prioritizing adults aged forty to fifty-nine years offers the most balanced and robustly cost-effective strategy in Korea under these constraints.

Marcus: They interpret the advantage of S2 in terms of model dynamics, suggesting that allocating doses to adults closer to higher fatality ages can avert substantially more deaths per dose than focusing primarily on younger adults.

Yuki: This contrasts with scenario S3 which focused on broader vaccination of adults aged twenty to fifty-nine years, but those gains required a much higher investment.

Ines: The limitations they noted include that long-term projections depend on assumptions about future transmission intensity captured through the force of infection, and they didn't explicitly stratify high-risk subgroups with markedly elevated fatality.

Marcus: These results underscore the importance of dynamic, context-specific economic evaluations for adult immunization policy in aging populations.

Yuki: Cost-effectiveness of Adult Hepatitis A Vaccination Strategies in Korea Under an Aging Susceptibility Profile is a study that grounds theoretical modeling in measurable clinical change for public health decisions.

Lucky paper: 2607.17405: Ines: We’re looking at Evaluating Conformal Reliability of Pathway-Level Transcriptomic Signatures Under Cross-Cohort Shift today.

Marcus: This paper looks at how to evaluate sepsis mortality prediction when you move the model to a new clinical setting with different patient groups.

Yuki: The researchers systematically compared gene level, pathway level, and hybrid molecular representations using four whole blood transcriptomic cohorts containing nine hundred thirty-six patients and two hundred forty-eight mortality events.

Ines: So they harmonized that data into a shared feature space with seven thousand six hundred sixty genes plus three different ways to represent the data.

Marcus: They then tested these representations using logistic regression, random forest, XGBoost, and LightGBM for each representation.

Yuki: The gene level and hybrid representations generally showed the strongest raw discrimination performance in those initial tests.

Ines: But pathway level representations actually showed greater robustness across those different model families.

Marcus: Pathway aggregation reduces dimensionality but might sacrifice some predictive resolution, which is a trade-off we need to consider.

Yuki: The paper introduces a Pathway Stability Index or PSI to quantify stable transferable signals that generalize across independent cohorts versus associations unique to one dataset.

Ines: A high PSI pathway, defined by criteria (three), had considerably greater external reproducibility and higher selection frequency.

Marcus: That suggests the study isn't just looking at AUROC or AUPRC anymore when we think about deployment.

Yuki: Exactly maximizing AUROC or AUPRC alone is insufficient when developing models intended for deployment across independent clinical settings because of representation choice.

Ines: They link those high-PSI genes back to Gene Ontology and KEGG enrichment analysis to see if they match known sepsis pathophysiology mechanisms like immune response.

Marcus: It shows that PSI is most useful when you pair it with functional enrichment and domain knowledge, not just the numbers.

Yuki: The study concludes that molecular representation influences not only predictive performance but also the reliability, interpretability, and transferability of transcriptomic prognostic models.

Ines: And they did flag a limitation: there’s still considerable heterogeneity among cohort sizes and transcriptomic platforms.

Marcus: Also, the conformal guarantees decrease as the extent of distribution shift increases in those settings.

Yuki: Additional validation will be required to ensure that PSI performs similarly across larger and more diverse datasets.

Lucky paper: 1504.05884: Ines: We're looking at Impact of density-dependent migration flows on epidemic outbreaks in heterogeneous metapopulations today because we want to see how movement changes where an outbreak starts.

Marcus: This paper uses a SIS-diffusion model and makes it nonlinear by making the movement rate depend on the population of the departure patch.

Yuki: The spatial arrangement is described by terms like p(k) for connectivity distribution and P(k'k) for how a patch of degree k connects to one of degree k'.

Ines: They then use those equations to describe the epidemic dynamics using the average number of susceptible and infectious individuals in each patch.

Marcus: The total number of individuals in the whole metapopulation is conserved, which they state as S(t) + I(t) = N rho zero where rho zero is that average number of individuals per patch.

Yuki: Density dependence comes into play through the diffusion rates for susceptible and infectious individuals, denoted as D S(rho k) and D I(rho k).

Ines: When there are no epidemics, they find a disease-free equilibrium rho* k that satisfies D S(rho* k) rho* k = M k, where M is some constant.

Marcus: This results in a power-law distribution for the patches, which they define as those with connectivity larger than k* which is calculated as D S(rho zero) rho zero / M.

Yuki: The initial epidemic growth rate is found using the dominant eigenvalue lambda one of the Jacobian matrix of equation (two) at that disease-free equilibrium.

Ines: For a constant contact rate, c(rho) is just c zero and that dominant eigenvalue equals zero beta c zero -, which they say is independent of the migration pattern.

Marcus: But when the contact rate becomes density dependent, a sufficient condition for lambda one to be greater than zero is that k

beta(c(rho* k) -) - (one - P(kk))D I(rho* k): > zero.

Yuki: Numerical computations show that the initial growth rate decreases as the migration exponent increases.

Ines: Strengthening migration from heavily populated areas can help contain an epidemic by causing it to appear in less populated areas, where the basic reproduction number is significantly smaller.

Marcus: For large enough migration exponents, they found that the onset of an epidemic may not be triggered by infectious individuals living in large size populations.

Yuki: This happens when individuals, also the infectious ones, have a high propensity to move from heavily populated areas.

Ines: Such a higher number of infectious individuals arriving at patches with lower population sizes makes it more likely for an outbreak to occur there.

Marcus: The percentage of population living in high-population patches always lies between thirty-six point seven nine percent and one hundred percent.

Yuki: The paper discusses distinct physical mechanisms behind the scaling laws in the topological features of mobility networks.

Ines: They suggest that for many insects and other animals, only local environmental conditions lead to density dependencies at the origin location for random dispersal within a metapopulation.

Marcus: But humans migrate with knowledge of political, economic, and environmental conditions in their potential arrival patches.

Yuki: Under that circumstance we can assume that the migratory flow between two patches depends on features of those involved patches.

Ines: The main novelty here is showing that migration patterns determine where epidemic outbreaks can take place with higher probability.

Marcus: That's a big difference from just assuming local dispersal rules apply everywhere.

Yuki: The paper also touches on how cooperation in structured populations can be promoted by conditional migration, allowing cooperators to escape areas rich in defectors.

Ines: This concept of context-dependent movement is relevant when considering stochastic viability in island models where dispersal is partial.

Marcus: We need to make sure we are looking at the numbers here, especially the migration exponent that causes that change in growth rate.

Lucky paper: 2509.02139: Ines: We are moving into our fifth and final paper of this session, and it’s a deep dive into how visual systems actually work by looking at the primary visual cortex, specifically the paper titled On sources to variabilities of simple cells in the primary visual cortex: A principled theory for the interaction between geometric image transformations and receptive field responses.

Marcus: It’s Tony Lindeberg’s work from KTH Royal Institute of Technology that we’re covering now.

Yuki: The core idea here is establishing an identity between visual system responses obtained from different observations under varying viewing conditions.

Ines: So the central thesis is that the family of receptive fields must possess specific covariance properties under certain classes of geometric image transformations to allow for formal matching of responses.

Marcus: What are these geometric transformations that they are testing?

Yuki: They investigate four main classes: uniform spatial scaling transformations, spatial affine transformations, Galilean transformations, and temporal scaling transformations.

Ines: The authors rigorously demonstrate that for idealized models based on Gaussian derivatives, the receptive field responses exhibit provable covariance under compositions of these four geometric image transformations.

Marcus: That sounds like a big theoretical requirement. What is the main prediction coming from this covariance?

Yuki: The overall prediction is that simple cells in the primary visual cortex of higher mammals ought to have receptive field shapes that are expanded over the degrees of freedom associated with these four main types of geometric image transformations.

Ines: So it’s not just fitting the data, but requiring a specific expansion based on the geometry of observation.

Marcus: It suggests that the structure of those geometric transformations must be reflected in the desired structure of receptive field responses.

Yuki: That leads to several specific hypotheses about observed variabilities, like spatial scaling and image rotations and elongation.

Ines: The paper also suggests that if a sufficiently general learning-based paradigm is exposed to varied image structures, the learned receptive fields should exhibit variabilities consistent with these theoretical predictions derived from geometric covariance.

Marcus: That connects the theory to actual learning paradigms, which is important for how we build AI models.

Yuki: And they also present an alternative modeling approach where visual receptive fields could explicitly represent only the finest spatial and temporal scales for each point in the visual field at each time moment.

Ines: So there seems to be a dichotomy there between expansion across all degrees of freedom and just representing the finest scales.

Marcus: That's interesting, it suggests a hierarchy or different ways systems encode information.

Yuki: This work provides a powerful theoretical basis for understanding how biological vision systems are structured to process complex, real-world visual data.