AI lost to a quantum computer: Moderna and IBM simulated the longest molecule

Researchers from Moderna and IBM successfully simulated the longest mRNA molecule using a quantum computer.

mRNA is a molecule responsible for transferring genetic information from DNA to ribosomes. It is used to create vaccines and control protein synthesis in cells. This molecule consists of a single chain of amino acids and has a secondary protein structure consisting of a series of folds.

With each added nucleotide, the number of folding options for the molecule increases. Therefore, predicting the shape of the molecule becomes difficult on a large scale.

Experiments by Moderna and IBM

Back in 2024, Moderna and IBM conducted research showing how quantum computing could be used for prediction. Initially, these predictions were based on AI models and classical binary computers.

New research has shown that it is possible to process mRNA sequences with thousands of nucleotides, but without complex elements. For example, “pseudoknots” have more complex bends and more complex internal interactions. By excluding such elements, the accuracy of protein folding prediction has certain limitations.

Researchers are working to overcome existing limitations, as predicting all details of the structure directly affects the development of effective vaccines and provides more reliable predictions.

A large number of experiments were conducted with quantum modeling algorithms. Initially, 80 qubits were used on the R2 Heron processor. This allowed scientists to predict the secondary structure of a protein 60 nucleotides long.

Previously, the best model consisted of 42 nucleotides. The increase in the experiment was possible due to the elimination of noise. Research was also conducted using 354 qubits in the absence of noise.

Increasing the number of qubits should significantly affect the accuracy of modeling and the ability to create longer sequences.