This article presents a sample web application for running protein structure prediction at scale using AlphaFold2 on AWS, combining a user-friendly frontend with a cloud HPC backend.

• Researchers can submit protein sequences via simple web interface without AWS console access
• Backend uses AWS ParallelCluster for flexible HPC cluster management and job scheduling
• FASTA files stored in S3, processed through API Gateway and Lambda functions
• AlphaFold2 databases hosted on Amazon FSx for Lustre for rapid cluster access
• Job history tracked in Amazon Aurora Serverless v1
• Entire infrastructure deployed using AWS CDK for automated setup
• Recurring costs approximately $290/month; per-job costs depend on GPU instance type
• Architecture supports alternative protein prediction software like OpenFold and OmegaFold

This solution simplifies protein structure prediction for researchers while enabling IT admins to manage scalable, secure computing resources through a single web interface.