Title

AWS re:Invent 2022 - Optimized software for improved performance in cryo-EM and genomics (LFS305-S)

Summary

Mike McManus from Intel introduces the session, highlighting cryo-electron microscopy (cryo-EM) and its importance in structural biology.
Sudej Kumar from Clovertex discusses the company's focus on high-performance computing (HPC) in drug discovery and healthcare, and their success in reducing drug discovery time.
Clovertex's platform focuses on structural biology and cryo-EM, with plans to expand to genomics and multi-omics.
Benchmarking of AWS instances for cryo-EM workloads is presented, showing cost and performance benefits of using optimized instances.
Professor Toshio from KEK shares insights on supporting researchers with cryo-EM data analysis and the benefits of using AWS for computational resources.
Brendan from Cention discusses the company's bioinformatics software for next-generation sequencing (NGS) data processing, which is sequencer agnostic and optimized for CPUs.
The session emphasizes the importance of choosing the right compute resources for cost and performance optimization in scientific research.

Cryo-EM is a critical technology for understanding the structure of proteins, and software tools are essential for converting blurry images into three-dimensional models.
Clovertex specializes in HPC services for the pharmaceutical industry, providing a platform that bridges research and IT, allowing scientists to focus on science rather than infrastructure complexities.
Benchmarking different AWS instances for specific workloads can lead to significant cost savings and performance gains, as demonstrated by Clovertex's analysis of cryo-EM workloads.
KEK's use of AWS and collaboration with Intel has led to improved processing costs and times for cryo-EM data analysis, highlighting the benefits of cloud computing in research.
Cention's bioinformatics software is designed to be efficient on CPUs, offering a cost-effective alternative to GPUs for NGS data processing, and is adaptable to various sequencer outputs.
The discussions underscore the trend towards cloud computing in scientific research, driven by the need for flexible, scalable, and cost-effective computational resources.
The session highlights the ongoing need for optimization and benchmarking in the deployment of scientific workloads on cloud platforms to ensure efficient use of resources and cost management.