How AI Could Change Biology

Sugars

Amino acids

Fatty acids

Nucleic acids

Its color

Its size

Its shape

Its weight

Proteins are too small to be observed.

The sequence of amino acids was unknown.

Individual parts of amino acids interact in many complex ways, making folding hard to predict.

There were no computers powerful enough to model proteins.

RoseTTAFold

AlphaFold2

DeepBlue

Watson

To make proteins stronger.

To help healthy cells grow faster.

To prevent harmful proteins from working.

To create new types of proteins.

Store large amounts of physical objects.

Mimic how biological neurons send signals.

Generate random numbers for scientific experiments.

Create new types of chemicals.

By guessing randomly until it gets it right.

By being programmed with every possible image.

By checking its predictions against known data and adjusting.

By asking humans for the correct answer each time.

AlphaFold is faster and uses less computing power.

RosettaTTFold can handle protein complexes, which AlphaFold struggles with.

AlphaFold is used for image recognition, while RosettaTTFold is for protein folding.

RosettaTTFold requires more human input than AlphaFold.