Reproduction of the CLAIRE de novo protein-binder design pipeline
A computational project — independent work alongside my wet-lab dissertation. I reproduced the CLAIRE small-molecule binder-design pipeline end-to-end, then pushed it onto two targets it had never been run on.
Three targets, not one
| Target | Why this one |
|---|---|
| Progesterone | the repository’s built-in example — a correctness baseline |
| Thalidomide | generalization test |
| 4-Hydroxytamoxifen (4-OHT) | generalization test |
The latter two were chosen deliberately: their polar-handle configurations are chemically distinct enough to probe generalization rather than repeat a known result.
All three produced AlphaFold2-self-consistent candidates — best case pLDDT 94.1 at 0.62 Å RMSD.
Why it matters to my own work
My dissertation designs bioPROTACs, which need a binding domain against their target — so the binder is the limiting reagent, and borrowing one confines you to targets somebody has already solved. Designing binders directly is how that constraint comes off.
Related: testing the generalizability of a multistate design method on KRAS
I extracted the computational recipe of Guo et al. (Science, 2025) from the primary literature and built a three-layer framework to test whether it transfers to the Ras superfamily:
- Calibration — T35S, a known state-switching mutant: does the diagnostic detect what it claims to?
- Clinical noise — G12D, G12C: does it fire on any substitution?
- Application — the hypervariable-region mutation from my own dissertation.
The first two layers are what make the third worth believing. This also meant rebuilding the simulation system in AMBER after diagnosing that a prior model was nucleotide-free — the replacement is GppNHp- and Mg^2+^-bound, with de novo parameterization for the ligand and the farnesylated cysteine anchor.
Toolchain
Rosetta/PyRosetta · ProteinMPNN · AlphaFold2/ColabFold · GROMACS · AMBER/AmberTools · OpenMM · CHARMM-GUI · antechamber, AM1-BCC, GAFF2 · Python · Linux and cloud GPU
I am Muzi Ouyang (欧阳木子) — I go by Mia. I am a PhD candidate in Chemistry at Tsinghua University, advised by Prof. Ying Tan (谭英) at the Institute of Biopharmaceutical and Health Engineering (iBHE) in Shenzhen, where I spend most of my time thinking about why protein degraders sometimes stop working, and what that failure can teach us.
My research sits at the intersection of targeted protein degradation, cancer drug resistance, and molecular dynamics. It started with an observation that didn’t make sense: a KRAS-targeting bioPROTAC I had engineered worked beautifully in one breast cancer cell line, and almost completely failed in its drug-resistant counterpart. Chasing that discrepancy led me to a mutation in a part of KRAS nobody had looked at closely before, and eventually, through all-atom MD simulations and a carefully designed control experiment, to a mechanism: the degrader’s activity depended on how the target protein clusters itself at the membrane, not just on how tightly it binds. That turned a puzzling failure into a diagnostic principle I now use to think about degrader design more generally.
From there, my work branched in two directions I care about equally. One is building degraders as actual therapeutic strategies, like a dual-target bioPROTAC that reverses tamoxifen resistance in breast cancer by co-degrading two proteins at once. The other is computational: I taught myself to run a full de novo protein design pipeline (Rosetta, ProteinMPNN, AlphaFold2) from scratch and used it to design new binders for small-molecule targets, which I’ve open-sourced on GitHub for anyone who wants to build on it.
What I’m looking for next is a postdoc where I can keep working at this intersection, bringing mechanistic, structure-aware thinking to new degradation chemistries and new disease contexts.