Harvard Medical School's Walter Lab runs 1.7 million protein interaction predictions using NVIDIA DGX Cloud
Harvard Medical School (The Walter Lab) · United States
The Walter Lab at Harvard Medical School, with access to NVIDIA DGX Cloud through the National Science Foundation's NAIRR pilot program, completed nearly 1.7 million protein-protein interaction predictions on ColabFold in three months using 32-node DGX clusters with 256 A100 GPUs, a task that previously would have taken years. The team curated 40,000 high-confidence protein interactions among 300 human genome maintenance proteins using two custom ML tools (KIRC and SPOC), publishing results via the Predictomes website.
Overview
The Walter Lab at Harvard Medical School, with access to NVIDIA DGX Cloud through the National Science Foundation's NAIRR pilot program, completed nearly 1.7 million protein-protein interaction predictions on ColabFold in three months using 32-node DGX clusters with 256 A100 GPUs, a task that previously would have taken years. The team curated 40,000 high-confidence protein interactions among 300 human genome maintenance proteins using two custom ML tools (KIRC and SPOC), publishing results via the Predictomes website.
The challenge
Other AI tools are trained to predict 3D structures of proteins but often do not reliably separate relevant protein-protein interactions (PPIs) from false-positive predictions, and manually determining protein structure was an incredibly time-consuming process that used to require years of research. The Walter Lab sought to expand its success with AlphaFold beyond a single lab's focus to the scale of the entire human proteome — 20,000 proteins interacting in perhaps a million ways.
The solution
The Walter Lab developed a multi-step pipeline: a machine learning tool called KIRC to identify protein pairs worth in-depth analysis by AlphaFold, and a second tool, the Structure Predictions and Omics-Informed Classifier (SPOC), to separate true and false PPI predictions. They first applied SPOC to a matrix of around 300 human genome maintenance proteins, generating 40,000 protein interaction predictions. To scale the project, the team adopted NVIDIA Accelerated Computing infrastructure and NVIDIA DGX Cloud — 32-node DGX clusters with 256 A100 GPUs — made available through NVIDIA's partnership with the National Science Foundation's NAIRR pilot program, running 1.7 million PPI predictions on ColabFold in three months. Results were published via the Predictomes website as an accessible, classifier-curated interactome resource.
Reported business value
The Walter Lab completed nearly 1.7 million protein-protein interaction predictions in three months — a process that would otherwise have taken years, if not more, according to graduate student Ernst Schmid. The classifier generated 40,000 high-confidence protein interaction predictions among 300 human genome maintenance proteins, providing new structural hypotheses for disease-relevant pathways.
Sources
Open any source and check the claim yourself — that is the point of the register.
Other education entries in the register.
Open Universities Australia triples lead qualification with generative AI agents
Open Universities Australia (OUA) used LivePerson's AI Studio to build an LLM-powered generative AI agent that handles initial prospective-student inquiries, replicating the conversational style of human advisors. The agent achieved 3x lead qualification rates compared to self-searching students and 2x compared to OUA's prior scripted chatbot, with an average response time of 6.3 seconds versus roughly 2 minutes for human advisors.
Miami Dade College boosts student pass rates by 15% with Microsoft 365 Copilot
Miami Dade College deployed Microsoft 365 Copilot among its President's Cabinet and senior leaders, and used Microsoft Copilot Studio to build AI-powered assistants supporting personalized learning, coding exercises, essay structures and math problems around the clock. In the first semester of implementation in STEM and advanced analytics courses, the college saw a 15% increase in student pass rates and a 12% decrease in dropout rates. A survey found 77% of users completed tasks faster, 76% saw improved work quality, and 81% reported increased productivity; if just 15% of MDC's more than 6,000 employees saved 12 minutes a day, that would total nearly 50,000 hours a year. Encouraged by these results, MDC expanded its Copilot deployment by 400 licenses, prioritizing student-facing employees.
GoGuardian: Safer schools, empowered teachers, thriving students
GoGuardian, which powers safe, focused learning for half of U.S. K-12 students, migrated its ML infrastructure to Databricks to manage billions of daily inferences for web filtering, classroom management and harm prevention while maintaining a PII-free, COPPA/FERPA-compliant data environment. Using Delta Lake, Lakeflow, Unity Catalog, MLflow and Databricks Model Serving, GoGuardian achieved up to 50% reduction in machine learning operational costs, 90% operational cost savings with its Delphi website classification model, and a 62% reduction in inappropriate device use among students. AI-driven prioritization also cut the volume of records requiring human review for high-risk content by over 95%, from 1 million to 35,000-45,000.
Lola - Success story at the University of Murcia
The University of Murcia deployed the chatbot Lola, built by 1MillionBot, to handle student queries during registration and university entrance exam periods. Lola has logged over 74,000 unique conversations with about 25,000 students, with a 93.4% accuracy rate. In June-July 2019, Lola answered queries from 5,525 users across 12,247 conversations and 34,423 messages, resolving 94.2% of queries, versus 28,000 inquiries handled by staff the year before implementation.
Was this helpful?
Your feedback helps us improve our use case database