Joint Neural Posterior Estimation and Model Selection for Chemical Reaction Networks

Abstract

Simulation-based Bayesian inference for Chemical Reaction Networks (CRNs) must simultaneously solve two computationally demanding problems: discriminating between overlapping reaction pathways and estimating reaction rate constants. Classical approaches, such as Sequential Monte Carlo ABC, rely on handcrafted summary statistics and scale poorly across large parameter spaces. To overcome this, we introduce a neural inference architecture. Our pipeline utilizes shared 1D-CNNs and a permutation-invariant Set Attention mechanism to learn low-dimensional summary statistics from variable-length, multi-experiment time-series data. These statistics simultaneously condition a model selection network to compute posterior model probabilities and model-specific conditional Invertible Neural Networks (cINNs) that estimate parameter posterior distributions. We demonstrate the practical utility of our approach on a real protodeboronation example, illustrating how it handles structural ambiguity and extracts kinetic parameters from competing catalyst deactivation pathways.

Mutual Linearity of Complexes in Chemical Reaction Networks
Pedro E. Harunari, Guilherme Fiusa, Matteo Polettini
arXiv8 Oct 2026PreprintCC BY 4.0Full textPhysical chemistry Statistical & nonlinear physicsarXiv
Optimal Control of Chemical Reaction Networks with Thermodynamic Cost
Tetsuya J. Kobayashi, Alix Robin, Shuhei A. Horiguchi
arXiv1 Nov 2026PreprintCC BY 4.0Full textPhysical chemistry Mathematical physicsarXiv
Large-deviations theory for growing chemical reaction networks
Praful Gagrani, Ignacio Madrid, Tetsuya J Kobayashi
arXiv25 Sep 2026PreprintCC BY 4.0Full textSystems & synthetic biology Physical chemistryarXiv
Generative discovery of chemical reaction networks
Yunhong Han, Gaopeng Ren, Yan Liu +16
arXiv28 Oct 2026PreprintCC BY-NC-SA 4.0Full textSystems & synthetic biology Physical chemistryarXiv
Hierarchical models for large chemical reaction networks
Jeremie Unterberger, Ulysse Herbach, Roxane Cellier
HAL24 Jun 2026PreprintCC BY-NC-ND 4.0Full textTheoretical & computational chemistry StatisticsHAL

Cite

BibTeX RIS

BibTeX

@misc{esquivel2026joint,
  title = {{Joint Neural Posterior Estimation and Model Selection for Chemical Reaction Networks}},
  author = {J. Arturo Esquivel and Jorge A. Campos-Gonzalez-Angulo and Sergio Pablo-García and Alán Aspuru-Guzik and Vianey Leos-Barajas},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.18825},
  eprint = {2610.18825},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.18825},
}

RIS

TY  - UNPB
TI  - Joint Neural Posterior Estimation and Model Selection for Chemical Reaction Networks
AU  - J. Arturo Esquivel
AU  - Jorge A. Campos-Gonzalez-Angulo
AU  - Sergio Pablo-García
AU  - Alán Aspuru-Guzik
AU  - Vianey Leos-Barajas
PY  - 2026
DA  - 2026/10/13
PB  - arXiv
DO  - 10.48550/arxiv.2610.18825
UR  - https://arxiv.org/abs/2610.18825
AB  - Simulation-based Bayesian inference for Chemical Reaction Networks (CRNs) must simultaneously solve two computationally demanding problems: discriminating between overlapping reaction pathways and estimating reaction rate constants. Classical approaches, such as Sequential Monte Carlo ABC, rely on handcrafted summary statistics and scale poorly across large parameter spaces. To overcome this, we introduce a neural inference architecture. Our pipeline utilizes shared 1D-CNNs and a permutation-invariant Set Attention mechanism to learn low-dimensional summary statistics from variable-length, multi-experiment time-series data. These statistics simultaneously condition a model selection network to compute posterior model probabilities and model-specific conditional Invertible Neural Networks (cINNs) that estimate parameter posterior distributions. We demonstrate the practical utility of our approach on a real protodeboronation example, illustrating how it handles structural ambiguity and extracts kinetic parameters from competing catalyst deactivation pathways.
ER  -

Abstract

Related

Cite