The development of artificial intelligence (AI) has been a huge focus for the past few years, but beyond generating lazy students’ essays and fooling thousands with generated images, artificial intelligence is extremely valuable in many fields, including biotechnology.

Proteins, the macromolecules that DNA codes for, have complex structure based on the building blocks they’re made of, amino acids, and how they interact with each other. Knowing the structure and function of proteins is extremely valuable because of their countless applications in biology, but they can consist of thousands of amino acids, making their bonds and interactions with other molecules extremely difficult to predict. Scientists have determined the structure of many proteins already, but very few are simple to predict while also having significant applications. Understanding more complex proteins could unlock new pharmaceuticals, industrial products, vaccines, and more. 

The AI program AlphaFold2 has made huge waves in understanding protein structure, as it can even render 3D models in extreme accuracy. In addition to structure, AI rendering can predict proteins’ function and how they interact with other molecules, which would be extremely useful for drug development. Scientists have grappled with predicting proteins’ 3D structure for decades, but in just a few years, AlphaFold2 has predicted the structure of almost every protein known to humans, over 200 million. AlphaFold3, released in 2024, can generate the structure of DNA, RNA, and some ligands, in addition to proteins, and offers breaking insights into proteins’ interactions with each other and with other molecules. Best of all, its technology is completely free to all scientists doing public research.

One team at Google Deepmind, the company that developed AlphaFold, is working to use this technology to address antibacterial resistance. As cases of antibacterial resistance climb, it is more important than ever to study its mechanisms. If bacteria continue to evolve as they have, diseases can become impossible to treat, and suddenly the common cold is a serious threat. Google Deepmind is exploring one way bacteria have rendered drugs ineffective, by altering their plasma membranes. By studying the enzymes involved in this process, they can gain a better understanding of how to address it. Unfortunately, proteins involved in cell membranes are difficult to isolate via traditional methods for determining structure, like X-ray crystallography. AI has expedited this process, opening the doors to confront antibacterial resistance in less time and with fewer resources.

However, AlphaFold still has room for improvement. Testing simulations of complexes between antigens and molecules such as T cell receptors and antibodies resulted in relatively low accuracy. Of course, this technology is in its infancy, and there is plenty of time for it to develop and expand.

Other companies are also using AI to expedite biological discoveries. There are thousands of diseases and approved drugs in the world, and drug discovery is costly and time intensive. Since many diseases attack the body through common mechanisms, researchers at Every Cure have considered changing existing drugs for different diseases, which can save a great deal of time and resources. Using AI helps match diseases to approved drugs way more efficiently than doing it manually, considering there are 4,000 drugs on the market and 18,500 known diseases. EveryCure will use AI to estimate the strength of each combination, then try to repurpose the best matches. The company believes it will create “generic repurposed treatments for 15-25 diseases by 2030” using this method, without having to discover any new compounds and test them for safety.

Another development in AI with a biological purpose is XLuminA, which can be used to optimize microscopic perspectives on a revolutionary scale. With microscopy, there are thousands of different optical rotations available, with different positions of mirrors, lenses and XLuminA can identify ideal frameworks to enhance the resolution of microscopic images. XLuminA “evaluate potential designs 10,000 times faster than traditional computational methods.”

AI is so new that it’s hard to say what will come next. AI is currently being considered to analyze brain signals in patients with mental disorders to prescribe the most effective medication for them, or interpret behaviors in addicts to better understand their process and risks. This field is growing fast, and each advancement opens doors to countless possibilities.

You really can’t talk about AI without addressing its downsides. AI poses a major threat to job security, intellectual property rights, and honestly the mental abilities of those that rely on it, but as with all technology, the problem is how it’s used, not the tool itself. In science, there are so many applications to AI that should be celebrated rather than overshadowed by its unrealized negative potential. I really respect that AlphaFold is entirely public, and I think we should spend more of our attention on celebrating that and holding other AI companies to similar standards. Hopefully advancements in technology will continue to lower the barrier to entry through saved time and costs, as AlphaFold has, instead of raising it, as other technologies seem to be interested in.

I think I may just be a tech optimist. With the introduction of the “dire wolves” (that are really grey wolves, as Hank Green has so eloquently pointed out), I can’t help but thinking, “that’s really cool though!” I mean, I’m not super keen on gene editing animals for funsies, I know they’re trying to say something about ecological niches, but it seems more like an excuse than their actual reasoning. I’m not against it or anything, I just think that funding could be used on gene editing technologies with medical or agricultural applications, you know, things that can help people. But then again, those funds could have also been used on fracking technologies, so hooray they were used on advancing the field of gene editing! I don’t know, it’s nuanced. But let’s give some credit where it’s due! And stay tuned to learn with me!

https://biotechnology.georgetown.edu/news/the-future-of-artificial-intelligence-and-biotechnology/

AlphaFold 2: https://www.nature.com/articles/s41392-023-01381-z 

AlphaFold 3: https://blog.google/technology/ai/google-deepmind-isomorphic-alphafold-3-ai-model/#future-cell-biology 

Antibiotic resistance: https://www.youtube.com/watch?v=uLDud7pNiNQ&list=PLqYmG7hTraZAhkAh72kzzLC4r2O4VoVgz&index=4 

Every Cure: https://www.audaciousproject.org/grantees/every-cure 

XLuminA: https://scitechdaily.com/10000x-faster-ai-discovers-new-microscopy-techniques-in-record-time/ 

https://scitechdaily.com/10000x-faster-ai-discovers-new-microscopy-techniques-in-record-time/