How Proteros accelerates drug discovery by using AWS ParallelCluster

This article describes how Proteros, a structure-based drug discovery company, implemented a scalable HPC solution using AWS ParallelCluster to accelerate Cryo-EM and Protein Crystallography processing.

• Proteros faced challenges processing terabytes of data per sample with fluctuating computational demands
• Built secure foundation using AWS Control Tower and hybrid connectivity to on-premises data center
• Deployed AWS ParallelCluster with multiple Slurm queues for elastic scaling and flexible resource allocation
• Implemented hybrid storage combining Amazon FSx for Lustre and Mountpoint for Amazon S3
• Reduced cloud storage costs by 60-70% compared to Lustre-only approach
• Integrated Active Directory for multi-user access and used Amazon DCV for remote visualization
• Achieved scalability, cost efficiency, operational automation, and improved security compliance

Proteros successfully modernized their HPC infrastructure on AWS, enabling scalable drug discovery processing with significant cost savings and maintained security standards.