cv
Contact Information
| Name | Tianyu Wu |
| Professional Title | Ph.D. Candidate, Biophysics and Quantitative Biology |
| tianyu16@illinois.edu | |
| Website | https://forxhunter.github.io |
Experience
-
Graduate Researcher, AI for Biology and Protein Design
Diwakar Shukla Group, University of Illinois Urbana-Champaign
- Learned ranker for protein-ligand pose prediction that reaches 86.0% top-1 accuracy on PoseBusters v2.
- Protein language models (ESM2) adapted with LoRA fine-tuning.
- Tests of whether LLM representations and outputs can be trusted, using acceptance guards and null controls (NeurIPS 2026 XAI4Science workshop).
- Methods for automatic scientific discovery: deciding which measurement separates competing mechanistic explanations of the same data.
-
Graduate Researcher, Whole-Cell Modeling
Luthey-Schulten Group, University of Illinois Urbana-Champaign
- Built the first 4D hybrid RDME-ODE simulations of the yeast galactose switch on the GPU-based Lattice Microbes platform, with cell geometry taken from cryo-electron tomograms.
- Benchmarked the multi-GPU solver under different spatial decompositions.
- Platform talk and session co-chair, Systems Biophysics, Biophysical Society Annual Meeting 2026.
Projects
-
2026 – Present Computational Protein Engineering: Form II Rubisco (Rhodospirillum rubrum)
An end-to-end computational pipeline for engineering a carbon-fixing enzyme, from predictive models to designs a collaborating wet lab can test.
- Trained and benchmarked sequence- and structure-based models on deep mutational scanning data; established calibration and error floors before using predictions to rank candidates.
- Ran molecular dynamics, free-energy, and quantum-chemical calculations on the catalytic site; built decoy and radial controls to test whether a proposed structural descriptor is actually site-local.
- Designed de novo binders and redesigned active sites with generative backbone models; converted the surviving designs into ordered constructs, matched controls, and assay protocols for experimental validation.
-
Predicting Protein Aggregation Propensity with Protein Language Models
A sequence-based machine-learning framework for aggregation propensity, from whole-protein prediction to residue-level aggregation-prone regions. First-author manuscript in preparation.
- Built a protein language model-based predictor of aggregation propensity, trained on a published high-throughput stability library and extended to residue-level aggregation-prone region detection.
- Designed nested and external validation protocols; benchmarked against public curated aggregation databases and training-free biophysical baselines.
- Ran in-silico deep mutational scanning on an antibody Fab; coordinated experimental validation (DLS, BLI) with collaborators.
-
Sep 2026 – Present MetaCarto
Automatic, deterministic layout of genome-scale metabolic models as Escher-compatible maps, with a browsable viewer for all 108 BiGG models.
- Metabolic networks
- Graph layout
- Open source
-
Oct 2026 – Present Can a virtual cell predict a pathway it has never seen switch on?
A pilot on 24,562 human blood cells that separates what expression data, mechanistic knowledge and a handful of stimulated cells each contribute to predicting an unseen interferon response.
- Virtual cell
- Single-cell perturbation prediction
-
Nov 2026 – Present Domain Module Contract
A proposal for shipping domain expertise as a module that plugs into each new foundation model instead of a fork of an old one.
- Foundation models
- Modular fine-tuning
Education
-
Ph.D. in Biophysics and Quantitative Biology
University of Illinois Urbana-Champaign
Selected Publications
-
Sep 2026 MetaCarto: biologically faithful automatic layout for genome-scale metabolic maps
bioRxiv
- T. Wu. An automatic layout framework that draws genome-scale metabolic models as curator-style pathway maps. Applied to all 108 BiGG models to produce 2,621 maps without manual rearrangement.
-
Feb 2026 Spatial heterogeneity alters the dynamics of the yeast galactose switch: Insights from 4D RDME-ODE hybrid simulations (meeting abstract)
Biophysical Journal 125(4), 15a
- T. Wu, M.-C. Spindler, A. Apsley, H. Li, Z. R. Thornburg, J. Mahamid, Z. Luthey-Schulten. Platform talk, Systems Biophysics, BPS 2026 Annual Meeting.
-
Jul 2025 Spatial Heterogeneity Alters the Dynamics of the Yeast Galactose Switch: Insights from 4D RDME-ODE Hybrid Simulations
bioRxiv
- T. Wu, M.-C. Spindler, E. Earnest, H. Li, Z. R. Thornburg, J. Mahamid, Z. Luthey-Schulten. The first 4D simulations of the galactose switch in a tomogram-based yeast cell.
-
Feb 2025 Visualizing Cell Structures with Minecraft
The Biophysicist
- T. Wu, Z. R. Thornburg, K. Tan, S. Kenkel, S. A. Boppart, R. Bhargava, Z. Luthey-Schulten.
Academic Service
-
Reviewer
International Conference on Machine Learning (ICML)
-
Reviewer
PLOS Computational Biology
Skills
Machine learning: Large language models, Protein language models (ESM2), LoRA fine-tuning, Learned ranking, Model evaluation
Scientific computing: GPU parallel programming, Lattice Microbes, Stochastic and deterministic simulation (RDME, ODE), Genome-scale metabolic modeling
Interests
Automatic scientific discovery: Hypothesis generation, Experiment selection, Identifiability of mechanisms
AI for Biology: Protein engineering, Aggregation propensity, Rubisco engineering, Protein design, Virtual cells