Baidu, a leading AI company with strong Internet foundation, and Stemirna Therapeutics, a leading mRNA biotechnology company, today announced the completion of pre-clinical studies that characterized the mRNA vaccine sequences for COVID-19 designed using Baidu’s proprietary algorithm LinearDesign. These studies indicated that the LinearDesign-based mRNA sequences significantly outperform the benchmark sequences designed by traditional algorithms in terms of stability, protein expression and immunogenicity, which are critical attributes of mRNA vaccines.

The results, published on the preprint server arXiv (https://arxiv.org/abs/2004.10177), validate the effectiveness of LinearDesign along with its promising value in the field of biopharmaceuticals.

mRNA vaccine has emerged as a ground-breaking vaccine platform for controlling and preventing infectious diseases, including COVID-19. However, despite its several beneficial features, such as rapid production and non-infectious properties, mRNA instability remains a major limitation that affects the storage, distribution, and efficiency of mRNA vaccines.

Since 2018, Baidu has been advancing computational biology research underlying LinearDesign, with a major effort devoted to the RNA secondary structure study. LinearDesign, first published on arXiv in April 2020 (https://arxiv.org/abs/2004.10177v4), has proved to be an efficient algorithm for optimized mRNA sequence design. Computer simulations have indicated that it takes just 11 minutes to use this novel algorithm to design an optimal mRNA sequence that can theoretically improve the stability and protein expression of an mRNA vaccine candidate.

To further validate the effectiveness of LinearDesign, Baidu teamed up with Stemirna Therapeutics, the first innovative company in China focused on the R&D of mRNA vaccines and drugs, to test seven mRNA COVID-19 vaccine sequences designed using LinearDesign. An mRNA sequence (benchmark) designed using the widely-used codon optimization method was tested side-by-side for comparison.

An illustration of seven mRNA COVID-19 vaccine sequences designed using LinearDesign

Results show that the LinearDesign-based mRNA vaccine sequences for SARS-CoV-2 Spike protein outperform the benchmark sequences in stability, protein expression, and immunogenicity. As to neutralizing antibody titer - one of the most critical parameters of vaccine efficacy – the level reported from LinearDesign-based mRNA sequences exceeds that of the benchmark by up to 20 times, potentially leading to a significantly higher immune protection. In addition, binding antibody titer and antigen-specific T cell response generated by LinearDesign-based mRNA sequences showed a 23-fold and 6-fold increase compared to those of the benchmark group, respectively.

LinearDesign’s application in mRNA therapeutics extends beyond prophylactic vaccines for infectious diseases to cancer, rare diseases, and others. Its early success is a prime example of applying computational tools to drug research and development, reflecting a disruptive force to the existing drug discovery process. With the mission of making the complicated world simpler through technology, Baidu will continue contributions to the global public health cause and building a community with a shared future for humankind.

You can learn about the origin and technical details of the LinearDesign algorithm from this IEEE Spectrum editorial piece by Dr. Liang Huang, Distinguished Scientist at Baidu Research USA: https://spectrum.ieee.org/covid-19-vaccine-rna