AlphaFold screen maps nearly 2,900 likely interactions among human mitochondrial proteins

Nov 18, 2026
Texas A&M researchers screened more than 630,000 protein pairs and confirmed several of the predicted pairings in the lab.

Scientists at Texas A&M AgriLife Research used Google DeepMind's AlphaFold-Multimer to screen every possible pairing of about 1,100 human mitochondrial proteins, more than 630,000 combinations, and found 2,895 likely interactions, according to a study published in Nature Communications.

More than half of those interactions had never been reported before. Around two years of lab experiments, some with researchers at Washington University School of Medicine in St. Louis, confirmed several of the predicted pairings in yeast and human cells. The study also gave 85 previously uncharacterised proteins their first probable partners.

Mitochondria are the energy centres of human cells, and mutations in nearly 30% of mitochondrial proteins have been linked to diseases ranging from cancer, Parkinson's and Alzheimer's to rare inherited metabolic disorders. The function of many of these proteins is still unknown, and finding which proteins they interact with is one way to infer it. The predictions linked many disease-associated proteins to previously unknown biological pathways.

The team has made its predictions available as a free online resource, MitoMatch. "We are a traditional biochemistry and genetics lab, not an AI lab," said lead author Abhinav Swaminathan, according to Texas A&M AgriLife.

We hope you enjoyed this article

Consider subscribing to one of our newsletters like Life AI Weekly or Daily AI Brief.

Also, consider following us on social media:

Free newsletter

Life AI Weekly

Weekly coverage of AI applications in healthcare, drug development, biotechnology research, and genomics breakthroughs.

Whitepaper

Tensordyne Napier: What If One Rack Could Do the Work of Nine?

Tensordyne

This Tensordyne whitepaper presents Napier, an inference-focused AI processor and rack-scale system based on the company’s TDN Math logarithmic number system. It examines infrastructure requirements for large mixture-of-experts and agentic models, compares major inference architecture approaches, and details the TDN AIP processor, TDN72 pod, TDN Link fabric, and Napier Ultra configuration. The paper reports simulation-based performance, cost, and accuracy-validation results, including Tensordyne’s projected comparison of one Napier rack with a nine-rack Nvidia Rubin plus Groq deployment; the chip is reported as taped out and in fabrication.

Read more
Free, six days a week

Daily AI Brief: the AI news that matters, in your inbox.