Bioinformatics

6 | Feb | 2026
De novo Design of Functional Nucleic Acids of Aptamers
This is a framework to use nucleic acids language model to generate functional nucleic acids.
Zhiming Zhang et al.
Open Capsule
Published in Nature Computational Science

Chemistry

12 | Dec | 2025
PropMolFlow: Property-guided Molecule Generation with Geometry-Complete Flow Matching
Molecule generation is advancing rapidly in chemical discovery and drug design. Flow matching methods have recently set the state of the art (SOTA) in unconditional molecule generation, surpassing score-based diffusion models. However, diffusion models still lead in property-guided generation. In this work, we introduce PropMolFlow, a novel approach for property-guided molecule generation based on geometry-complete SE(3)-equivariant flow matching. Integrating five different property embedding methods with a Gaussian expansion of scalar properties, PropMolFlow achieves competitive performance against previous SOTA diffusion models in conditional molecule generation across various properties while preserving the stability and validity of the generated molecules, consistent with its unconditional counterpart. Additionally, it enables faster inference with significantly fewer time steps compared to baseline models. We highlight the importance of validating the properties of generated molecules through DFT calculations performed at the same level of theory as the training data. Specifically, our analysis identifies properties that require DFT validation and others where a pretrained SE(3) geometric vector perceptron regressors provide sufficiently accurate predictions on generated molecules. Furthermore, we introduce a new property metric to assess the model's ability to propose molecules with underrepresented property values, assessing its capacity for out-of-distribution generalization. Our findings reveal shortcomings in existing structural metrics, which mistakenly validate open-shell molecules or molecules with invalid valence-charge configurations, underscoring the need for improved evaluation frameworks. Overall, this work paves the way for developing targeted property-guided generation methods, enhancing the design of molecular generative models for diverse applications.
Cheng Zeng & Jirui Jin
Open Capsule
Published in Nature Computational Science

Bioinformatics

23 | Nov | 2025
Self-optimized Spectral Distance for Low-Light High-Throughput Raman Hyperspectral Imaging
The code for Self-optimized Spectral Distance for Low-Light High-Throughput Raman Hyperspectral Imaging
Yurong Chen
Open Capsule
Published in Nature Computational Science

Computer Science

10 | Oct | 2025
Revealing Neurocognitive and Behavioral Patterns by Self-Supervised Manifold Learning from Dynamic Brain Data
This repository contains the implementation of BCNE (Brain-dynamic Convolutional-Network-based Embedding), a self-supervised manifold learning method for analyzing dynamic brain data. BCNE is designed to reveal neurocognitive and behavioral patterns through dimensionality reduction and visualization of complex neural recordings.
Zixia Zhou
Open Capsule
Published in Nature Computational Science

Physics

29 | May | 2025
Iterative variational learning of committor-consistent transition pathways using artificial neural networks
Code and data used in "Iterative variational learning of committor-consistent transition pathways using artificial neural networks" by A. Megías, S. Contreras-Arredondo, C.-G. Chen, C. Tang, B. Roux and C. Chipot submitted to Nat. Comp. Sci. The file "run.py" is created to run one iteration for learning the committor model and the committor-consistent string via the PCCANN given a trajectory of a biased trajectory of a carbon atom under the action of the Müller--Brown potential.
Alberto Megías et al.
Open Capsule
Published in Nature Computational Science

Medical Sciences

12 | May | 2025
Virtual brain twins for stimulation in epilepsy (Part 2)
Estimating the epileptogenic zone network (EZN) is an important part of the diagnosis of drug-resistant focal epilepsy and plays a pivotal role in treatment and intervention. Virtual brain twins, based on personalized whole brain modeling, provide a formal method for personalized diagnosis and treatment. They integrate patient-specific brain topography with structural connectivity from anatomical neuroimaging such as MRI, and dynamic activity from functional recordings such as EEG and stereo-EEG (SEEG). Seizures demonstrate rich spatial and temporal features in functional recordings, which can be exploited to estimate the EZN. Stimulation-induced seizures can provide important and complementary information. We consider invasive SEEG stimulation as the most practical current approach, and temporal interference (TI) stimulation as a potential future approach for non-invasive diagnosis and treatment. This paper offers a virtual brain twin framework for EZN diagnosis based on stimulation-induced seizures. This framework estimates the EZN and validated the results on synthetic data with ground-truth. It provides an important methodological and conceptual basis for a series of ongoing scientific studies and clinical usage, which are specified in this paper. This framework also provides the necessary step to go from invasive to non-invasive diagnosis and treatment of drug-resistant focal epilepsy.
Huifang E Wang, Borana Dollomaja, Paul Triebkorn & Viktor Jirsa
Open Capsule
Published in Nature Computational Science

Medical Sciences

12 | May | 2025
Virtual brain twins for stimulation in epilepsy (Part 1)
Estimating the epileptogenic zone network (EZN) is an important part of the diagnosis of drug-resistant focal epilepsy and plays a pivotal role in treatment and intervention. Virtual brain twins, based on personalized whole brain modeling, provide a formal method for personalized diagnosis and treatment. They integrate patient-specific brain topography with structural connectivity from anatomical neuroimaging such as MRI, and dynamic activity from functional recordings such as EEG and stereo-EEG (SEEG). Seizures demonstrate rich spatial and temporal features in functional recordings, which can be exploited to estimate the EZN. Stimulation-induced seizures can provide important and complementary information. We consider invasive SEEG stimulation as the most practical current approach, and temporal interference (TI) stimulation as a potential future approach for non-invasive diagnosis and treatment. This paper offers a virtual brain twin framework for EZN diagnosis based on stimulation-induced seizures. This framework estimates the EZN and validated the results on synthetic data with ground-truth. It provides an important methodological and conceptual basis for a series of ongoing scientific studies and clinical usage, which are specified in this paper. This framework also provides the necessary step to go from invasive to non-invasive diagnosis and treatment of drug-resistant focal epilepsy.
Huifang E Wang, Borana Dollomaja & Paul Triebkorn
Open Capsule
Published in Nature Computational Science

Chemistry

24 | Apr | 2025
Machine Learning Many-Body Green's Functions
A code capsule for peer review of the manuscript titled "Unified Deep Learning Framework for Many-Body Quantum Chemistry via Green's Functions." The main code for this manuscript runs from code/mlgf, and depends on the publicly available fcmdft and libdmet codes also included in the capsule. The readme for mlgf can be found in its respective folder. To test the code, launch a jupyterlab (or notebook) session, find the example jupyter notebook under code/mlgf/examples, and run the code chunks while following the markdown instructions. The intention is to generate GWGWGW data for ammonia, methane, water, and ethane, train MBGF-Net on ammonia, methane, and water, then finally predict the MBGF of ethane and obtain several many-body properties.
Christian Venturella et al.
Open Capsule
Published in Nature Computational Science

Physics

13 | Mar | 2025
Bi-level Identification of Governing Equations for Nonlinear Physical Systems
The identification of governing equations from data for nonlinear physical systems has been a long-standing challenge in scientific discovery. However, the inherently ill-posed nature of this inverse problem often leads to false discoveries that overfit certain datasets without capturing the true dynamics of the studied system. To overcome the challenge, we propose a Bi-level Identification of Equations (BILLIE) framework in this work to simultaneously perform equation discovery and validation in two hierarchical objectives of a bi-level optimization. BILLIE is effectively solved using policy gradient techniques from reinforcement learning and hence offers the possibility of identifying very challenging physical systems where the existing approaches fail. Our experimental results on the Navier-Stokes equation, the Burgers' equation, and the three-body system demonstrate BILLIE's dominant performance over existing methods. Moreover, we apply BILLIE to the task of discovering the RNA and protein velocity equations from real-world single-cell sequencing. For the first time, we can quantitatively discover these prominent velocity equations in a data-driven manner, which marks a departure from previous reliance on experts' empirical intuitions to hypothesize these equations. Notably, the equations identified by BILLIE surpass empirical equations in accurately characterizing the future states of differentiated cells. This demonstrates BILLIE's strong potential to contribute to the discovery of fundamental physical rules underlying diverse scientific phenomena.
Zeyu Li et al.
Open Capsule
Published in Nature Computational Science

Social Sciences

19 | Mar | 2025
Gender and racial diversity socialization in science
Code to run null models for gender and racial diversity socialization.
Weihua Li, Hongwei Zheng, Jennie E. Brand & Aaron Clauset
Open Capsule
Published in Nature Computational Science