Adaptyv competitions act as a benchmark for protein design methods, using standardized wet-lab experiments to show how well their designs perform when they are actually made and tested in the real world.
Each competition sets a design challenge: what the protein should do and how its performance will be measured. Anyone can enter for free, using any design method. Adaptyv then synthesizes the selected designs and tests them in the same lab, under the same protocols, so the results can be compared directly.
All competition data is published openly, including results from designs that did not work.


Each competition has its own rules, selection process and timeline, but most follow the same four stages.
We announce the design challenge, the entry requirements and how the designs will be tested. Anyone can enter using any design method.
Requirements vary by competition. For EGFR and Nipah, designs had to be single-chain proteins under 250 amino acids and at least ten mutations away from any known binder.
We usually receive more designs than we can test, so the selection process varies by competition.
Nipah divided its 1,200 testing slots between computational ranking, expert review and a community vote. RBX1 distributed 300 slots evenly across the participating teams.
We work with organizations of all kinds, including universities, research institutes, foundations and AI labs. Partners can co-sponsor a competition and design the challenge with us.
You can bring a specific protein target, one or more functional goals, or a broader research question.
Together, we define the challenge, decide how entries will be evaluated and plan the wet-lab testing.


Adaptyv synthesizes the proteins and measures their performance. Every design follows the same experimental protocols, making the results directly comparable.
For binding competitions, we measure affinity using the Adaptyv standard protocol.
Every tested design is published on Proteinbase with its sequence, predicted structure, design method and experimental measurements—whether it worked or not.