Technology

Miniproteins

Miniproteins are very small and highly structured proteins.

AI Proteins designs miniproteins that combine the most important, drug-like features of small molecules and antibodies.

small molecule
(~0.2 kDa)

miniprotein
(3-9 kDa)

antibody
(~150 kDa)

Miniproteins constitute a new class of drug with the following properties:

  • stable in high temperatures, to harsh chemicals, and digestive enzymes
  • low cost to manufacture, ship and store
  • refrigeration not required in manufacturing, storage, or distribution
  • precise control over pharmacokinetics
  • produced using chemical or biological synthesis
  • tunable affinity and selectivity
  • lower risk of side effects due to unwanted binding or toxic breakdown products

AI proteins designs miniproteins de novo that can bind to any target using proprietary artificial intelligence.

Product Discovery Engine

AI Proteins uses proprietary artificial intelligence to design novel protein structures de novo. Learn more

Libraries are constructed and screened to find the best miniprotein binders. Learn more

Automated laboratory robotics are used to accelerate production, testing and optimization. Learn more

Miniprotein products are tested to ensure the posession of excellent biophysical properties. Learn more

Computational Design

Backbone Design

We begin designing miniproteins by crafting the underlying shape and structure, which we call the “backbone” of the molecule. This is accomplished by sequentially stitching together small fragments of known structure observed in nature.  As the structure grows, we sample lots of different fragments to find each piece that best fits the shape that we’re building.

Disulfide Bond Design

Disulfide bonds are covalent “molecular staples” that enhance the strength and durability of the miniprotein. We identify all possible locations in our backbone that are capable of hosting one of these bonds. Then, we select the best combination of bonds and incorporate them into the structure.

Amino Acid Design and Final Optimization

We identify a sequence of amino acids that will direct our backbone to fold into the designed shape. This is performed by sampling different chemical groups throughout the backbone until the best possible combination is identified. Finally, the whole molecule undergoes a refinement step to optimize the interplay between the backbone shape and amino acid chemistry.

Discovery of Miniproteins

Library Construction

To construct randomized libraries we diversify residues on a wide array of designed scaffolds to obtain libraries of millions of potential miniprotein binders.

We also construct targeted libraries where the interface is rationally designed to bind a specific epitope on a target of interest.

Library Screening

Display technologies allow us to find the needles in a haystack: the best miniprotein binder(s) in the constructed libraries.

Library Optimization

We use synthetic biology and additional diversification rounds to improve the binding of the selected miniproteins.

Production of Miniproteins

We use liquid-handling robots to express and purify a large number of miniproteins in parallel.

Characterization of Miniproteins

To minimize the failure in late development stages, we ensure that all our miniproteins possess excellent biophysical properties.

Correct protein folding

Durability

High affinity

High specificity

Proper oligomerization state