High-performance computing on the cloud facilitates gene sequencing

01 Introduction to Shenshi Technology

Shenshi Technology has the ability of scientific research and industrial implementation. The company's scientific research team consists of multiple fields such as physical modeling, numerical algorithms, machine learning, high-performance computing, and drug and material computing. The self-developed new-generation molecular simulation algorithm can perform quantum-mechanical-accurate computational simulations on billion-atom-scale systems on the basis of maintaining quantum-mechanical accuracy, combined with high-performance computing.

Shenshi Technology uses artificial intelligence and molecular simulation algorithms, combined with advanced computing methods to solve important scientific problems, and creates a new generation of micro-scale industrial design and simulation platform for the most basic biomedicine, energy, materials and information science and engineering research of human civilization.

02 E-HPC Best Practice of Shenzhen Trend Technology

Shenshi Technology Drug Design Platform is a new generation drug design platform based on artificial intelligence, physical modeling and high-performance computing. The platform provides complete solutions for various scenarios in the preclinical drug discovery stage, including structural modeling, target discovery, virtual screening, molecular recommendation, binding affinity assessment, and macromolecule drug design, and provides one-stop computing services.

The business needs of Shenshen Technology mainly include three points:

First, the business volume of Shenshi Technology fluctuates greatly, and it hopes to provide elastic scalability based on workload to maximize the use of computing resources.

Second, high-performance operation and maintenance management. Shenshi Technology has a large number of supercomputing clusters, and computing power resource scheduling and computing power resource management are rigid needs. Due to the heavy workload of the clusters, Shenshi Technology hopes to provide detailed cost bills for operations, which is convenient for billing.

Third, high cost performance. As a start-up company, Shenshen Technology pursues high cost performance and hopes to complete effective calculations at a relatively low cost.

With its leading products and technical support, Alibaba Cloud has become the choice of Shenshi Technology. First of all, to address the issue of elasticity, Alibaba Cloud's ultra-large-scale CPU, GPU, computing power resource pool and elastic supply capacity can respond to various computing needs of customers at any time.

Alibaba Cloud is one of the few cloud vendors in China that provides highly elastic and high-performance computing platform E-HPC on the cloud. The automatic scaling capability of the Alibaba Cloud E-HPC cluster surpasses the static resource allocation of traditional supercomputing. Users can expand or shrink capacity on demand according to the load.

Through custom settings, the job throughput can be effectively improved, the job processing speed can be accelerated, and the utilization rate of the cluster can be greatly improved.

Secondly, for efficient operation and maintenance management, Alibaba Cloud's automatic operation and maintenance and visual management technology, E-HPC, can provide customers with centralized management, job management, user management and visualization capabilities. While not changing the usage habits of scientific computing business, it provides more efficient elastic cluster scheduling service and makes better use of computing resources.

The Alibaba Cloud platform realizes active operation and maintenance of clusters. The applicability of hot upgrade, migration and automatic operation and maintenance makes Shenshen Technology no need to manage the life cycle of hardware resources and improves the efficiency of cluster management. Allow customers to focus on application and scientific research, and use the convenience brought by cloud supercomputing more conveniently.

At the same time, E-HPC provides an open-source billing tool based on the bill of the billing center, which supports detailed cost statistics of operations and efficiently solves billing needs.

To meet the needs of cost-effectiveness, the computing power resources on Alibaba Cloud have a flexible pricing model, which can be billed on an as-needed basis. Use more flexible purchase methods to comprehensively optimize computing power costs. Alibaba Cloud's solution meets the needs of Shenshi Technology to build a high-availability computing power cluster at low cost.

By adding Alibaba Cloud computing power clusters, the average synthesis scale of each pipeline is reduced from thousands to dozens, which greatly reduces the cost of synthesis and reduces the waiting time for custom synthesis.

Uni-Fold overcomes the limitations of AlphaFold2's non-open source training code, single hardware support, and non-commercial model. Adaptation, performance optimization, and function improvement on NVIDIA GPUs have been carried out in the training and inference links.

Hermite Uni-FEP can accurately predict the ability of protein-ligand binding. With the existing protein-ligand co-crystal structure, under the guidance of the experience of pharmaceutical chemists, the initial ligand molecule is modified and modified, and the affinity difference before and after modification is accurately calculated with the help of FEP, which replaces unnecessary experiments and achieves cost reduction and efficiency increase.

E-HPC implements multi-node parallel computing, and the job scheduler solves elastic management of computing resources and task scheduling.

03 Summary

Alibaba Cloud's high-performance computing provides the life science industry with massive elastic computing capabilities, convenient resource task management tools, efficient data storage and analysis reports, effectively supporting the needs of the life science industry and ensuring orderly and efficient business operations.
Alibaba Cloud and Shenshi Technology have jointly created a life science computing platform on the cloud with lower cost and better computing power to achieve inclusive computing power. Fusion goes deeper.