A new era of protein
designers.

Powering the future of purpose-built proteins.

Fable's innovative machine learning strategy harmonizes biology with computation, revolutionizing therapeutic development.

Our platform leverages extensive public and proprietary datasets housing billions of sequences, alongside protein structures, protein-protein interactions, all tied together through a sophisticated machine learning infrastructure. The result?

Precision-engineered biologics tailored for optimal potency, developability, binding affinity, and target engagement geometry.

The benefits of
purposeful design.

Our approach is designed to overcome current therapeutic index limitations and channel response, where and when it matters.

This capability empowers us to craft innovative medicines tailored to tackle complex biological challenges.

Alanine Parallax

A timeline of firsts.

Our team is at the forefront of applying artificial intelligence to protein design. Here are seminal publications from the past decade.

2016

First

to screen fully designed protein

libraries

(Science Adv. 2016)

to screen fully designed protein libraries

(Science Adv. 2016)

2017

First

protein backbone generative

model

(PLOS CB, 2017)

protein backbone generative model

(PLOS CB, 2017)

2019

First

graph neural network for

protein design

(bioRxiv 2019, Cell Syst 2020)

graph neural network for protein design

(bioRxiv 2019, Cell Syst 2020)

2020

First

pre-trained graph transformer

for peptide binding site prediction

(bioRxiv 2020, Commun Biol 2022)

pre-trained graph transformer for peptide binding site prediction

(bioRxiv 2020, Commun Biol 2022)

2021

First

to show LLM embeddings improve

mutation effect prediction

(JMB 2021)

to show LLM embeddings improve mutation effect prediction

(JMB 2021)

2022

First

validated protein backbone

diffusion model

(bioRxiv 2022, Nature Comp Sci 2023)

validated protein backbone diffusion model

(bioRxiv 2022, Nature Comp Sci 2023)

2023

First

hierarchical transformer;

LLM to solve ZF design

(Nature Biotech 2023)

hierarchical transformer; LLM to solve ZF design

(Nature Biotech 2023)

2024

First

Boltzman generator; generates peptide

conformational ensembles from sequence

(Nature Machine Intelligence 2024)

Boltzman generator; generates peptide conformational ensembles from sequence

(Nature Machine Intelligence 2024)

Precision platform
designed to solve
for complexity.

At the core, we utilize two proprietary foundation models: a structure-based model and a sequence-based model.

Through precision engineering at the atomic scale, our platform learns patterns within the 3-D structures and amino acid sequences of proteins and protein complexes.

Structure

Our structure model is pre-trained on a large set of data, built on state-of-the-art architectures and optimally reasoning in 3-D spaces.

Sequence

Our sequence model is pre-trained on billions of sequences, learning the languages of biologics.

De novo design

Our continuous-time diffusion model generates novel binders just based on target structure.

optimization

Our platform enables rapid generation of novel designs with optimized potency and developability properties.

“The pace of progress in the
field is astounding.
At Fable, we remain at
the forefront of advancing
cutting-edge methods,
models, and tools.”

Philip M. Kim, CTO

Philip M. Kim

Setting our
platform apart.

Our 7 key differentiators drive precise protein design, seamlessly integrating binding and developability criteria into lead candidates.

1

Unique data strategy overcomes antibody-antigen scarcity

2

Tight integration of experimental data through active learning

3

Structure-based generative model built on a global frame equivariant transformer

4

Integration of powerful sequence-based and structure-based models

5

Fully template-free de novo conditional generation of binders

6

Rapid co-optimization of multiple developability parameters and binding

7

Designed to deliver diverse, potent and selective binders with best-in-class developability

“Powering an ML platform
to deliver better medicines
requires an unrelenting
focus at the bench to
continuously generate data
of the highest quality.”

Vanita D. Sood, SVP,
Head of Drug Discovery

Vanita D. Sood

We've assembled a team with diverse and specialized skill sets, uniquely equipping us to explore the fast-paced world of protein design and drive forward new possibilities in medicine.

Our Team

Arginine Parallax