Work · Biochemistry

Things I built at the bench.

Five years of synthetic microbiology and enzyme chemistry — from protein NMR in Zürich to cell-free carbon fixation and mRNA at Academia Sinica under James C. Liao (廖俊智). A wall of the projects, big and small.

Cell-free CO₂ fixation

A molecular printer for CO₂

Academia Sinica · Liao Lab

A cell-free, self-replenishing enzyme cycle that fixes CO₂ into C2–C4 products outside any cell, sustained six hours at rates that beat photosynthesis. I hand-purified roughly half the enzymes and ran the assays that proved the cycle turns over.

Succinate, malate, pyruvate and glyoxylate — built out of CO₂, not sugar
Oxygen-insensitive, self-replenishing rGPS + MCG cycle
¹³C-bicarbonate MS/MS tracing to prove self-replenishment
Plus a whole track of my work that never made the paper

Succinate
Malate
Pyruvate
Glyoxylate
Protein purification
¹³C MS/MS
Enzyme assays
Cell-free

Read the report Nature Catalysis, 2022 ↗

AGG U Ψ U Ψ CCGGGAGG U Ψ C U Ψ ACGCCAGCACACA U Ψ G U Ψ GAGCA U Ψ CCGA U Ψ CGA U Ψ C U Ψ U Ψ U Ψ AC U Ψ U Ψ GGG U Ψ CCAGA G U Ψ G U Ψ U Ψ AG U Ψ ACC U Ψ AG U Ψ G GGA U Ψ GA U Ψ CGGC U Ψ C U Ψ U Ψ G U Ψ CACGCAGCGGGGCCA C U Ψ ACGGGCG U Ψ AA U Ψ C U Ψ U Ψ AGG U Ψ U Ψ AACAAG U Ψ U Ψ A U Ψ C AAC U Ψ AGAG U Ψ ACAGCCA GAA U Ψ CACC U Ψ U Ψ A U Ψ U Ψ GGA GGGACGCCCGCG U Ψ GGC A U Ψ CGCAAAA U Ψ GG U Ψ U Ψ CA CCCCG U Ψ G U Ψ GAG U Ψ ACGC U Ψ GCCGACGAGAA U Ψ AGA U Ψ A U Ψ CGA U Ψ G U Ψ AGG U Ψ G U Ψ U Ψ A U Ψ U Ψ U Ψ CAAGGC U Ψ AGCCG U Ψ GCGAA U Ψ G U Ψ GAGCGG U Ψ GACCGAAAGCC U Ψ CCAG A U Ψ U Ψ ACCC U Ψ G U Ψ ACGAA U Ψ GC U Ψ C U Ψ ACAAGGG U Ψ CC U Ψ A U Ψ U Ψ U Ψ GCGA U Ψ CGGCCGG

mRNA chemistry

mRNA + pseudouridine

COVID-era · Liao Lab

A long, hands-on program around N1-methylpseudouridine (N1-met-Ψ) — the modification that makes mRNA vaccines stable, but which was expensive and poorly understood. We pushed on it from several directions at once.

Convert every uridine to N1-met-Ψ after transcription (reverse of the Moderna / Pfizer route)
Flip a single motif-specific uridine via guide-RNA + H/ACA ribonucleoprotein
Make the modified nucleotide biochemically instead of by costly chemical synthesis

RNA modification
guideRNA
H/ACA RNP
Enzyme synthesis

Carbon-fixing enzymes

Pyruvate synthase for CO₂

Liao Lab

Exploring pyruvate synthase (PFOR) and pyruvate formate-lyase as alternative entry points for feeding carbon and electrons into central metabolism — routes to fix CO₂ where the main pathway wouldn’t reach.

PFOR
PFL
Ferredoxin

Directed evolution

Bacterial evolution for precision fermentation

Liao Lab

Adaptive laboratory evolution to turn fragile lab strains into stable production strains. Directed evolution is powerful but leaves hyper-mutating, unstable cells; the real work was engineering genomic stability for industrial-scale fermentation — quantified with rifampicin fluctuation assays.

ALE
Strain stability
Fluctuation assay

Synthetic pathways

NOG strain evolution

Liao Lab

Engineering and evolving the non-oxidative glycolysis (NOG) pathway in E. coli, pushing carbon yields toward — and past — 100%. I replayed the strain’s evolution to see which mutations recur; colonies kept converging on the same insertion-sequence hits in ptsG.

NOG
Genome evolution
E. coli

Synthetic methylotrophy

SM1 methylotroph evolution

Liao Lab

Evolving SM1 — a synthetic methylotroph, an E. coli rewired to grow on one-carbon feedstock. I ran the adaptive laboratory evolution myself, pushing the strain’s growth and stability on C1 substrate. (Results from these runs to be added here.)

Synthetic methylotrophy
ALE
C1 metabolism

Computational biology

Codon optimizer

Derived from the mRNA work

A spin-off from the mRNA project: a Python codon-optimization algorithm that scores on di-codon (codon-pair) frequencies rather than single codons. The pairings carry signal standard tables throw away — and they matter for expression and drug efficacy.

Python
Codon optimization
Algorithms

Coursework · CS & statistics

A year of theory

TIGP · first year

The doctoral program front-loaded a year of theory — and the statistics training was exceptional, well past what I strictly needed. Algorithms from divide-and-conquer to Smith–Waterman, computational methods in medicine, and a rotation in the lab of Mark Liao (a YOLOv4 co-author), where I got my first real footing in deep learning.

Statistics
Algorithms
Deep learning

Genomics

DNA sequencing

Liao Lab

A lot of hands-on sequencing — reading genomes to verify constructs, track mutations through evolution experiments, and close the loop between the strains I built and what actually changed in their DNA.

Genome sequencing
Variant analysis

Foundations · Zürich

Protein NMR & metal-ion binding

UZH · MSc

Where it started — a Master’s under Oliver Zerbe, studying proteins by NMR and how bound metal ions reshape them. Also my first real programming, in R and Bash.

NMR
Structural biology

Notebook

…and more

There’s more than fits on one wall — mini-projects, assays, and side quests I’ve half-forgotten. This card is a placeholder: tell me what’s missing and it goes up here.

.wall { grid-auto-rows: auto !important; align-items: start; } .wall .tile { grid-row: auto !important; } .wall .tile-card { position: relative !important; height: auto !important; } .wall .tile-fill { flex: none !important; } .wall .tile-blurb { display: block !important; } .wall .tile-highlights, .wall .tile-tags { display: flex !important; max-height: none !important; }

The way I work now — thinking in systems, data, and structure — was forged here.

Why search needs rethinking The text-to-graph project