software

Open-source software and platforms developed by Madduri Lab for privacy-preserving federated learning, genomics analysis, and scientific workflows.

Our lab develops open-source software and platforms that advance privacy-preserving machine learning, large-scale genomics analysis, and scientific workflow automation. Below are our major software projects.

APPFL - Argonne Privacy Preserving Federated Learning

An open-source framework for privacy-preserving federated learning that enables collaborative AI model development without centralizing sensitive data. **Key Features:** - Differential privacy and secure aggregation - Support for heterogeneous computing environments (cloud, HPC, edge) - Flexible communication backends (MPI, gRPC) - Policy-driven governance and access control - Integration with PyTorch and other ML frameworks **Applications:** Healthcare AI, genomics, smart grid optimization, COVID prediction models
APPFL/APPFL APPFL/APPFL

APPFLx - Federated Learning as a Service

A production-ready platform that provides privacy-preserving federated learning capabilities as a managed service.

Highlights:

  • Deployed on AWS with enterprise-grade security
  • Used by NIH-funded Bridge2AI program
  • Supports cross-institutional collaborations
  • Web-based interface for non-expert users
  • End-to-end encryption and secure enclaves

Access: appflx.link

MapperTrac

A massively parallel, portable, and reproducible tractography pipeline for neuroimaging analysis.

Features:

  • Scalable across HPC clusters
  • Reproducible containerized workflows
  • Published in Neuroinformatics (2024)
LLNL/mappertrac LLNL/mappertrac

Globus Genomics

A next-generation sequencing analysis service built on Galaxy, Globus, and Amazon Web Services that democratized large-scale genomics analysis.

Impact:

  • Used by thousands of researchers worldwide
  • Analyzed millions of genomes
  • Led to successful commercial spinoff (funded by UChicago Polsky Center, NIH, NSF SBIR)
  • Established patterns for Science-as-a-Service platforms
globusgenomics/genomics-galaxy globusgenomics/genomics-galaxy

Historical Projects

caGrid - Cancer Biomedical Informatics Grid

Grid infrastructure for secure data sharing and high-performance workflows in cancer research, adopted by multiple NCI-designated cancer centers.

View on GitHub

Reliable File Transfer (RFT)

The first “grid service” created for Wide Area Networks, part of Globus Toolkit versions 3 and 4. Integral component for data staging in/out of HPC resources, adopted by thousands of institutions worldwide.

Getting Started

Interested in using our software? Here are some resources:

We welcome contributions and collaborations. See individual repositories for contribution guidelines.