Real Science, Real Workflows: How Researchers Are Using ACCESS Tools

Wondering how other researchers are actually using ACCESS tools in their workflows? The new Tools in Practice: Researcher Stories series on ACCESS Support does exactly that — short, detailed profiles of real projects showing how tools like Open OnDemand, Globus, Pegasus, and XDMoD fit into working research pipelines.

The first six stories span a wide range of fields and resource types:

• A chemistry team used Open OnDemand and Globus across Expanse and Bridges‑2 to run large‑scale simulations that revealed — for the first time at the atomic level — how PFAS "forever chemicals" break down under electrical potential
• An agricultural AI project used Globus for automated drone imagery transfer and Open OnDemand to make GPU‑accelerated machine learning accessible to interdisciplinary teams without HPC backgrounds
• A computational biology student screened 600,000 natural compounds for potential monkeypox treatments using Pegasus on the OSG Open Science Pool, then validated top candidates with molecular dynamics simulations on Delta GPU
• A fluid dynamics team compiled and ran custom HPC libraries on Bridges‑2 to simulate how soft tube deformability controls bubble behavior — findings with applications from soft robotics to inkjet printing
• A CMU machine learning team trained large language models on NCSA DeltaAI, using XDMoD to monitor GPU efficiency in real time — a discovery that directly shaped how they optimized their training pipeline
• A structural biology facility at Texas A&M used Globus to automate terabyte‑scale CryoEM data transfers to ACES, cutting time‑to‑first‑results from days to hours and enabling near‑atomic resolution protein structure determination

Each story includes the researcher's project summary, software used, allocation details, and practical advice for others considering a similar workflow.

More stories are in development. If you'd like to share how you're using ACCESS tools, lissie [at] sweetandfizzy.com (subject: Researcher%20Stories) (we'd love to hear from you).