Breakthroughs often happen through chance encounters. The stars line up: someone’s hunch meets another person’s skills and sparks fly. And just maybe – eureka.
Is there a way to make that reaction happen more often? It’s a question that’s been thrown around by biologists and machine learning experts at InstaDeep, especially when the world was locked down and looking to science for something groundbreaking – and fast.
Well, pep met peptides and those discussions over virtual coffees became in-person catch-ups and led to the DeepChain Apps Hub. A virtual watering hole where ideas from biologists and machine learning experts can cross-pollinate. It’s a place where a biologist might share a data set, or a machine learning expert might share an AI app focused on a protein problem.
The heart of the DeepChain Apps Hub is InstaDeep’s open source software that allows anyone with a laptop and an internet connection to quickly build AI protein Apps. To help, InstaDeep provides protein embeddings, high definition representations, generated by powerful transformer models that have trained with billions of proteins to make it faster and easier to create personalised machine learning models. The team is also offering the valuable datasets, as well as ready-to-use DeepChain Apps, including those made by the community.
Before Deepchain Apps, to work on protein research you needed massive computing resources, ML models and more, often making it out of reach for all but the most elite institutions. InstaDeep hopes that by democratising access to innovative AI, we can all benefit.
To learn more, visit deepchainapps.com
