We will accelerate the development of life science innovation

In early December, "Melting supercomputing Momentum, Creating a New chapter of Simulation - 2021 Supercomputing Innovation Application Conference" was held in Guangzhou. The conference adopted the way of "main forum +4 parallel sub-forums", and the report content was wonderful. Among them, the "Life and health" sub-forum attracted the attention of many participants. Around the national "Healthy China 2030" strategy, the integration and innovation of supercomputing, big data and artificial intelligence continue to enable the development of cutting-edge technologies in life and health. Let's review the wonderful content of the sub-forum together.

▲ Life and Health sub-forum

The "Life and Health" sub-forum was chaired by Professor Yang Yuedong, Deputy chief engineer of the National Supercomputing Guangzhou Center of Sun Yat-sen University, and Engineer Chen Pin of the High Performance Computing Department. Experts, scholars and industry elites from universities, research institutes and enterprises were invited to discuss the latest application practice and development trend of high-performance computing, big data and artificial intelligence integration and innovation technology in the field of life and health.

1. AI accelerates life and health science research

With the development of artificial intelligence technology, researchers rely on supercomputing for RNA structure prediction, pathological diagnosis, and new drug research and development, which has brought new breakthroughs in related fields.

▲RNA structure prediction

The fundamental reason for the seemingly endless discovery of new Rnas is that each genome has a huge library of non-coding Rnas with unknown functions. Although more and more non-coding Rnas have been found to play key roles in various life processes and to be involved in many diseases, the vast majority remain unknown. Zhou Yaoqi, a senior researcher from Shenzhen Bay Laboratory, brought a report titled "RNA: The Never-ending Frontier report introduced the challenges faced by deep learning in RNA structure prediction, including how to find reliable homologous sequences to better capture sequence information, how to improve the accuracy of secondary structure prediction through deep learning, etc. Zhou Yaoqi, a senior researcher, conducted an in-depth analysis on these issues, and mentioned that for the problem of too little data, Transfer learning can be used to approximate the secondary structure data and add 3D structure sequence information to increase the sequence information and improve the effect. In using sequence prediction RNA, adding artificially generated sequences can replace natural sequences to improve accuracy.

▲ Pathological diagnosis

"Pathology is the foundation of medicine", pathology is the gold standard of clinical diagnosis, traditional clinical diagnosis is faced with complex diagnostic process, diagnosis cycle is too long, automation level is not high and over-dependence on pathologists and other phenomena, therefore, "digital pathology +AI" (that is, computational pathology) technology came into being. Associate Professor Yu Jinguang of South China University of Technology introduced the concept and development history of computational pathology in a simple way in the report "The Foundation of Artificial Intelligence Enabling Medicine", and summarized and summarized a series of problems faced by current computational pathology research institutes, such as huge data volume, strong data heterogeneity, and high cost of data annotation. At the same time, Professor Yu also shared some typical cases of applying deep learning to pathological diagnosis research. Finally, Professor Yu said that the research of computational pathology has just started, and the analysis of the microenvironment of cancer cells and the research of multi-center multi-modal algorithms may become a hot spot in the future.

2. Large-scale computing helps solve life science problems

The powerful computing power of supercomputers has injected new momentum into the discovery of new targets and the development of innovative drugs. Xu Yong, a researcher from Guangzhou Institute of Biomedicine and Health, Chinese Academy of Sciences, presented a report entitled "Application of High performance Computing in anti-tumor drug Research". In response to the bottleneck problem of the current shortage of new targets for original drugs, researcher Xu Yong introduced a method to explore original innovative targets from clinical samples. Using computational biology and molecular simulation as means, the team carried out research on new targets and new drugs for prostate cancer by relying on supercomputing. Aiming at the characteristics of key androgen receptor proteins in the treatment of prostate cancer, the research team analyzed the characteristics of potentially effective small molecule drugs, used algorithms to screen out tens of thousands of molecular libraries from hundreds of millions of molecules, and further screened out molecules with protein target specificity and suitable for clinical use. After layers of screening and optimization, the research team finally obtained a drug with high anti-tumor activity, providing an important target and drug candidate for overcoming clinical resistance.