arXiv subjects

Wandi Xu

Publications and source records attributed to Wandi Xu.

4 recordsLinked to original sources

arXiv2026

Binding Affinity between Polymer Dots (Pdots) and Ovalbumin Protein at Varying pH

Determining the binding mechanisms between polymer dots and proteins is important for developing novel nanotechnologies for biomedicine and bioimaging. In this work, we use all-atom molecular dynamics simulations to determine the binding affinity of Pdots with ovalbumin protein at pH = 7 and 1. The selected Pdots are mixtures of Poly[(9,9-dioctylfluorenyl-2,7-diyl)-alt-co-(1,4-benzo-(2,1',3)-thiadiazole)] (PFBT) and poly(styrene/maleic anhydride) (PSMA) with varying composition. At pH = 7, the Pdots have a net negative charge due to the COO- functional groups on the PFBT, and the protein also has a net negative charge. At pH = 1, the Pdots are charge neutral with PFBT containing only COOH functional groups, and the protein also has a net positive charge. We sample the initial position of the protein by varying its initial position through all 6 orientations of a cube. For each orientation, we pull the protein towards the PFBT region of the Pdot. We compare the Coulombic and Lennard-Jones interaction energies for the 6 different interacting faces and two selected pH values. We find that the LJ interaction energies are similar among all 12 of these cases. The measured Coulombic interaction energies suggest that pH = 1 has better binding affinity than pH = 7. The potentials of mean force (PMF) along the pulling coordinates differ with pH. The PMFs from 2 of the 6 initial configurations at pH = 1 are negative whereas none of them are negative at pH = 7, confirming the preferred binding affinity when pH = 1. One of the faces at pH = 1 has the lowest PMF of about -30 kcal/mol, which is much lower than about 6 kcal/mol seen for the lowest case at pH = 7. Comparison of protein residue charge distributions at pH = 7 and 1 further shows that the electrostatic interaction is critical to the binding affinity, and negatively charged residues reduce at pH = 7 does not bind to negatively charged Pdot.

physics.chem-ph↗

arXiv2025

RAPTOR-GEN: RApid PosTeriOR GENerator for Bayesian Learning in Biomanufacturing

Biopharmaceutical manufacturing is vital to public health but lacks the agility for rapid, on-demand production of biotherapeutics due to the complexity and variability of bioprocesses. To overcome this, we introduce RApid PosTeriOR GENerator (RAPTOR-GEN), a mechanism-informed Bayesian learning framework designed to accelerate intelligent digital twin development from sparse and heterogeneous experimental data. This framework is built on a multi-scale probabilistic knowledge graph (pKG), formulated as a stochastic differential equation (SDE)-based foundational model that captures the nonlinear dynamics of bioprocesses. RAPTOR-GEN consists of two ingredients: (i) an interpretable metamodel integrating linear noise approximation (LNA) that exploits the structural information of bioprocessing mechanisms and a sequential learning strategy to fuse heterogeneous and sparse data, enabling inference of latent state variables and explicit approximation of the intractable likelihood function; and (ii) an efficient Bayesian posterior sampling method that utilizes Langevin diffusion (LD) to accelerate posterior exploration by exploiting the gradients of the derived likelihood. It generalizes the LNA approach to circumvent the challenge of step size selection, facilitating robust learning of mechanistic parameters with provable finite-sample performance guarantees. We develop a fast and robust RAPTOR-GEN algorithm with controllable error. Numerical experiments demonstrate its effectiveness in uncovering the underlying regulatory mechanisms of biomanufacturing processes.

stat.ML↗

arXiv2024

Linear Noise Approximation Assisted Bayesian Inference on Mechanistic Model of Partially Observed Stochastic Reaction Network

To support mechanism online learning and facilitate digital twin development for biomanufacturing processes, this paper develops an efficient Bayesian inference approach for partially observed enzymatic stochastic reaction network (SRN), a fundamental building block of multi-scale bioprocess mechanistic model. To tackle the critical challenges brought by the nonlinear stochastic differential equations (SDEs)-based mechanistic model with partially observed state and having measurement errors, an interpretable Bayesian updating linear noise approximation (LNA) metamodel, incorporating the structure information of the mechanistic model, is proposed to approximate the likelihood of observations. Then, an efficient posterior sampling approach is developed by utilizing the gradients of the derived likelihood to speed up the convergence of Markov Chain Monte Carlo (MCMC). The empirical study demonstrates that the proposed approach has a promising performance.

stat.ML↗

arXiv2023

Structure-Function Dynamics Hybrid Modeling: RNA Degradation

RNA structure and functional dynamics play fundamental roles in controlling biological systems. Molecular dynamics simulation, which can characterize interactions at an atomistic level, can advance the understanding on new drug discovery, manufacturing, and delivery mechanisms. However, it is computationally unattainable to support the development of a digital twin for enzymatic reaction network mechanism learning, and end-to-end bioprocess design and control. Thus, we create a hybrid ("mechanistic + machine learning") model characterizing the interdependence of RNA structure and functional dynamics from atomistic to macroscopic levels. To assess the proposed modeling strategy, in this paper, we consider RNA degradation which is a critical process in cellular biology that affects gene expression. The empirical study on RNA lifetime prediction demonstrates the promising performance of the proposed multi-scale bioprocess hybrid modeling strategy.

q-bio.MN↗