Daily Summary for 2026-10-27

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

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

Marcus: 4 new papers came out today.

Ines: 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 Detecting absence: A dedicated prediction-error signal emerging in the auditory thalamus.

The day's research: Ines: We are focusing on relaxation dynamics within chemical reaction networks today.

Marcus: These systems often show slow, plateau-like behaviors that matter biologically.

Yuki: We obtained bounds on the Kullback--Leibler divergence to equilibrium.

Ines: This involves mass-action chemical reaction networks with equilibrium.

Marcus: The bounds are characterized using singular values of the stoichiometric matrix.

Yuki: And time-integrated activities via deformed exponential functions <a href="/papers/2602.18932">AKMPC</a>.

Ines: This framework characterizes slow relaxation induced by local convexity in catalytic networks with plateaus.

Marcus: It offers a bound-based approach to quantifying these slow dynamics.

Ines: Beyond chemical kinetics, research looks at how the brain processes sensory information.

Marcus: Specifically focusing on detecting absence rather than presence seems key here.

Yuki: We tested the hypothesis that absence detection relies on a dedicated prediction error mechanism for updating beliefs about stimulus occurrence in the auditory pathway AKMPC.

Ines: Your fMRI studies showed feature prediction error is encoded in the auditory midbrain, but absence prediction error was not.

Marcus: That suggests absence prediction error is processed by a different circuit and these mechanisms might be disrupted in conditions like psychosis.

Yuki: So the difference lies in which circuit handles the absence signal?

Ines: Yes, that points toward distinct processing pathways for presence versus absence signals.

Ines: We modeled gene expression fluctuations using stochastic processes and rare events.

Marcus: That partitioning of Poisson arrivals framework helps model post-transcriptional regulation.

Yuki: This gives a general framework for large deviation rate functions for AKMPC.

Ines: It opens new avenues for analyzing rare events in general models of regulation.

Marcus: How does this apply to specific biological settings?

Yuki: We can analyze different settings using these new avenues.

Ines: Did the work yield concrete results regarding protein production rates?

Marcus: Yes, we derived bounds on the Kullback--Leibler divergence to equilibrium for AKMPC.

Yuki: That bound applies specifically to mass-action chemical reaction networks with equilibrium.

Ines: What about analyzing rare events in general models?

Marcus: The framework allows us to study rare events across various settings.

Yuki: We can use this approach for post-transcriptional regulation models.

Ines: Can we apply this to the prediction-error signal emerging in the auditory thalamus?

Marcus: That relates to Detecting absence, which asks how the brain knows what is out there.

Yuki: It provides a way to analyze that specific signal.

Ines: What about decomposing brain networks into gradient, curl, and harmonic flows?

Marcus: Hodge-Decomposition of Brain Networks analyzes networks by these three components.

Yuki: That decomposition helps us understand the structure of those networks.

Ines: Does this decomposition help with the large deviation objects we seek?

Marcus: It provides a structural tool for analyzing those large deviations.

Yuki: We create a general framework for objects of interest in post-transcriptional regulation.

Ines: That framework is conditioned on what specific condition?

Marcus: It is conditioned on the large deviations inherent in the models.

Yuki: The results depend entirely on those large deviation statistics.

Ines: So, we have methods for both rare events and regulation analysis.

Marcus: Indeed, these methods cover both areas effectively.

Yuki: That covers our review findings for today's material.

Ines: Detecting absence is worth hearing because it explores how the brain senses absence. Post-transcriptional Regulation of Stochastic Gene Expression Conditioned on Large Deviations is worth hearing as it provides a general framework for large deviations in regulation models.

Yuki: Convex Analysis of Relaxation Dynamics in Chemical Reaction Networks and Generalized Gradient Flows is worth hearing because we obtain bounds on the Kullback--Leibler divergence to equilibrium for mass-action chemical reaction networks with equilibrium.

Marcus: Hodge-Decomposition of Brain Networks is worth hearing because it analyzes brain networks by decomposing them into three orthogonal components: gradient, curl, and harmonic flows.

Ines: Those are the two papers we focus on today.

Lucky paper: 2511.21605: Ines: So, we are looking at Detecting absence: A dedicated prediction-error signal emerging in the auditory thalamus.

Marcus: The core idea is that detecting sensory absence relies on a specific type of prediction error called absence prediction error, or aPE.

Yuki: The authors hypothesize that this happens because feature prediction errors, which deal with pitch and perceived sound features, are not enough for updating beliefs about the world.

Ines: They argue those feature errors only model Gaussian distributions centered in the predicted and perceived pitch, which isn't sufficient for updating internal beliefs.

Marcus: So absence prediction error is paramount to this process because it updates our beliefs on stimulus occurrence itself.

Yuki: The study proposes that the separation between positive and negative prediction errors might also be a hallmark of these error signals responsible for the update of beliefs on occurrence.

Ines: How did they set up their experiment to isolate this absence prediction error?

Marcus: They developed a novel paradigm to elicit it, avoiding neural entrainment confounds by introducing different levels of expectations on the omissions.

Yuki: In that trial, participants heard a sequence of up to eight repetitions of a pure tone with an inter-stimulus-interval of five hundred ms.

Ines: The setup included one omission with probability p0 equals six over seven and that omission could occur in positions four five or six of the sequence.

Marcus: The expectations on those omissions were modulated by three abstract rules that the participants were shown.

Yuki: For instance, the probability of finding an omission in position four after hearing three tones was p0 over three while in position six after five tones it was p0.

Ines: What did they find regarding unexpected absence during the auditory pathway?

Marcus: They showed that unexpected absence occurred at all stages of the auditory pathway, and they verified that auditory regions adapted to the repetition of the standard tone.

Yuki: But responses to unexpected omission om4 were stronger than responses to repeated tones std1 in all auditory regions tested.

Ines: The significant responses actually appeared in bilateral PAC and MGB but not in IC at that stage.

Marcus: Specifically, the response to om4 showed a dependency with stimulus presence plausibility in MGB and PAC, while there was no detectable aPE signal in IC.

Yuki: When comparing contrasts normalized by BOLD sensitivity, the normalised omission om4 greater than std1 and the aPE contrast om4 greater than om6 were stronger in MGB than in IC.

Ines: That confirms that absence prediction error emerges along the ascending auditory pathway and doesn't seem to increase towards PAC.

Marcus: When we normalize by noise levels instead of BOLD sensitivity, similar results were obtained confirming that aPE emerges in MGB and does not seem to increase towards PAC.

Yuki: This finding suggests that the detection of absence is based on a specific type of prediction error computed by dedicated circuits.

Ines: The authors suggest this dissociation means there's a mechanistic separation between identifying a stimulus and identifying its features, which isn't present in feature identification because evidence for pitch is always direct.

Marcus: They also suggest that hallucinations typical of psychosis can be understood as the failure of perceptual systems to update the plausibility of incorrectly predicted stimuli from internal representations.

Yuki: Understanding this dissociated computation is essential for comprehending both perceptual function and dysfunction, especially in conditions like psychosis.

Ines: So, the conclusion is that IC only encodes presence prediction error, while MGB and PAC encode both presence and absence prediction error.

Marcus: That pattern of encoding is distinct from feature prediction error which we know is strongly encoded in IC.

Yuki: The results indicate that IC encodes only the presence prediction error, while MGB and PAC encode both presence and absence prediction error.

Ines: Detecting absence is based on a specific type of prediction error computed by dedicated circuits in the auditory system.

Lucky paper: 2607.17004: Ines: So we’re looking at Post-transcriptional Regulation of Stochastic Gene Expression Conditioned on Large Deviations, which is paper number two of two for this show today.

Marcus: This paper takes the Partitioning of Poisson Arrivals framework and maps general models onto promoter-based regulation models.

Yuki: They partition mRNA arrivals into N types, where each type corresponds to an individual partition. In the limit N goes to infinity, this mapping leads to a reduced model that looks like the Markovian arrival process or MAP model of transcription.

Ines: That reduced model's dynamics are described by a system of master equations given by ∂P(a, t)/∂t = D0P(a, t) + D1P(a − one t).

Marcus: The theory of large deviations quantifies the probability of observing large fluctuations for a specified trajectory observable in the steady-state.

Yuki: The probability distribution of the protein count in the full model Phi(A, t) can be constructed from that of each reduced model by Phi(A, t) = lim N→∞ X a1+a2+…+aN =A Y N j=one ϕj (aj, t).

Ines: The moment generating function G(λ, t) can also be represented in terms of the reduced model by G(λ, t) = X a e − λaϕ(a, t) N = ⟨e − λa⟩t N.

Marcus: The key insight from large deviation theory is that the dominant eigenvalue of D(λ), which controls the long-term behavior of the system, is-psi(λ).

Yuki: For sRNA repression models, they showed that protein burst sizes remain geometrically distributed from the unregulated model with a modified mean.

Ines: They also looked at how a given fold change R can be decomposed as R = ˜kp/kp ⟨T˜⟩/⟨T⟩, where ˜kp and ⟨T˜⟩ are the quantities that characterize the driven process for realizing R.

Marcus: When modeling mRNA decay as an arbitrary n-step process, the repetitive structure of the tilted generator for the reduced model yields a clean analytical derivation of the SCGF of the full model.

Yuki: The SCGF is given by Ψ(λ) = km / (one − µm / (µm + kp(one − e−λ))n. As n increases from one step Markovian case, the rate function monotonically narrows, implying that more degradation steps act as a buffer of stochastic gene expression by decreasing the probability of realizing rare protein production rates for both low and high activity levels.

Ines: So increasing n makes it harder to realize rare protein production rates in both scenarios.

Marcus: The marginal buffering ability of a degradation step is substantial when there are only a few steps, but as n increases, the marginal buffering ability rapidly monotonically decays to be negligible.

Yuki: The key advantage of this framework is that it allows for more detailed insights into the control of rare events by specifying the contribution of each individual parameter to the driven process.

Ines: But there's a limitation they mentioned: mRNA arrivals must follow a Poisson process, and they cannot incorporate promoter-based regulatory mechanisms like transcriptional bursting.

Marcus: That's a limitation because we can't model transcription initiation events directly in this framework.

Yuki: Nonetheless, the results obtained have opened new avenues for understanding rare events in gene expression modulated by post-transcriptional regulation.