01
Reason to understand
Integrate biological mechanisms, protein structure and cell state to expand how models reason about the biological world.
FUNCTION-FIRST PROTEIN DESIGN
Define function through reasoning.
Realize design through generation.
Therapeutic proteins, designed on demand.
Integrating biology, protein structure and cell state —
from understanding biology to designing therapeutic sequences.
LexicodonTHERAPEUTICS
01 / THE PARADIGM SHIFT
The next paradigm in protein design
starts with function.
02
Generate with purpose
Turn biological understanding into design criteria. Guide generation with reasoning to jointly optimize binding and function.
03
Learn through experiment
Make every round of validation the starting point for the next improvement in model performance.
02 / LEXIFORGE CLOSED-LOOP INTELLIGENCE
From cell state
to therapeutic sequence.
LexiForge connects protein generation, functional prediction and active learning in a continuously improving loop.
01 / GENERATE
Protein generation
Design therapeutic sequences by working backwards from cell state.
02 / EVALUATE
Functional prediction
Simulate cell fate to read out functional responses such as activation, killing and exhaustion.
03 / LEARN
Active learning
Use wet-lab data to continually calibrate models and inform the next round of design.
↺ Experimental feedback
Every validation advances the next round of design.
03 / VALIDATION & SERVICES
Validated design. Ready for your next discovery.
82.9%
Wet-lab hit rate
PD-L1 target · 29/35 tested
One-shot de novo miniprotein binders
>95%
Expression success rate
100% expression for PD-L1 miniproteins
Over 90% expression for VHHs
0.879 nM
Best binding affinity
One-shot de novo generation
PD-L1 Miniprotein binder Kd
85×
DRY-LAB HARD FILTER
Dry-lab filter pass rate for de novo candidates
Relative to other models, including PXDesign and RF3; computational screening metric reported in the one-pager.
CASE STUDY
IFNAR2 Binder
VHH
44% Hit rate · 156 nM Kd
CASE STUDY
PD-L1 Binder
Miniprotein · 100 AA
<1 nM Kd
OUR SERVICES
Bring validated capabilities to your research.
01 / DE NOVO BINDER DESIGN
De Novo Binder Design
Generate candidate binders through de novo sequence design for a specified target protein and binding site.
02 / AFFINITY ENGINEERING
Affinity Engineering
Use computational design and sequence optimization to increase or decrease the binding affinity of an existing protein for different applications.
04 / OUR NAME
Rooted in nature.
Written in sequence.
Our Chinese name, Lai Xi Cheng Xu, reflects a return to first principles and the laws of nature. “Cheng Xu” means forming a protein sequence and echoes the word for a computer program.
Protein function, programmed and refined.
05 / CAREERS
Help design the future of proteins.
We are seeking passionate candidates to join us in exploring AI-driven protein design and closed-loop experimentation.
OPEN ROLE
AI Protein Designer
To explore the role, send your CV and relevant project or research work.