AI is yet again making significant strides in the field of biology by accelerating evolution. It is making better CRISPR tools (Clustered Regularly Interspaced Short Palindromic Repeats) at an unprecedented pace. CRISPR is essentially a system that scientists borrowed and improved from bacteria. It acts as a programmable molecular scissor that can cut DNA at almost any place you choose. With the help of AI, the CRISPR technology is now being used to design new, synthetic versions of gene-editing enzymes that could be expanded or improved. It was observed that these artificial enzymes performed at par or better than the natural ones when tested in human cells.
To understand the significance of this breakthrough, we need to understand that DNA is the master blueprint for building and running life, and its precise editing requires targeting, cutting and rewriting. There are certain natural enzymes that do this, such as breaking food into smaller pieces or gluing two pieces together or copying or delivering instructions. You can think of these enzymes as hybrid molecular scissors, glue guns, copiers and delivery trucks. Now, genes are specific sections of that DNA master blueprint and contain recipes for making proteins. Because genes are long strings of DNA molecules, to change them, you need tools that work at the molecular level inside living cells. These tools are the enzymes and kind of act as a smart robot that can find a typo in a billion-page book and fix it, while the book is still being used.
The usual approach adopted by the researchers has been to tweak the existing enzymes using traditional CRISPR tools, but now the researchers are using AI protein design to generate entirely new and synthetic enzymes called SynTnpBs from scratch or with guided variation. The AI models have been able to predict sequences that would fold and bind correctly, recognize targets, and cleave DNA effectively. Many of these synthetic enzymes achieved 46-50% editing efficiency at a test gene in human cells.
This breakthrough is a logical step forward after the Nobel Prize-winning (2024) AlphaFold, an AI system designed by Demis Hassabis, CEO of Google DeepMind, with John Jumper, which predicted the 3D shape of any protein just from its amino acid sequence. Without AlphaFold, generating new synthetic enzymes would have been impractical.
The natural evolution process takes billions of years to create these families of enzymes, but AI compresses that into mere computational time and explores far beyond what nature provides. The trends are clearly pointing towards an era where soon AI will design novel biology, including new proteins and organisms.
There is a good chance that a significant number of diseases and ailments will simply get eradicated in the next few decades. AI combined with human-in-the-loop has reduced the cost of discoveries tremendously. What used to cost billions of dollars in yesteryears now costs only a small fraction of that. Key challenges such as good training data and significant human oversight still remain, yet the time compression is real, and we can expect more breakthroughs at an accelerated pace in the near future. This will only compound.