Projects

Two Senior RSEs provide support to research fellows in The Eric and Wendy Schmidt AI in Science Fellowship Programme. These fellows, who are conducting research on diverse topics across various departments in the MPLS division, can request assistance with software engineering, data analysis, and machine learning aspects of their work. In addition to offering long-term project development support, the RSEs also hold drop-in sessions for ad hoc assistance.

We are supporting a multi-institution the BBSRC grant BB/V018930/1 to maintain and enhance the functionality of Chaste, a mature open-source framework for the simulation of computational models in biology. This work is led in Oxford by Fergus Cooper, a Principal RSE with substantial experience in computational modelling, who works together with RSEs in Sheffield and Nottingham. Core components of the work include modernising the infrastructure, extending functionality by adding new cell-based modelling frameworks, increasing simulation speed by leveraging GPGPU offloading, and working to build the user community.

Global.health is an epidemiological platform established during the COVID-19 pandemic to provide open access to de-identified line list case data, facilitating research into infectious diseases and emerging outbreaks. The platform’s primary mission is to deliver timely and accurate data to decision-makers, researchers, and the public during the critical first 100 days of an outbreak. The development of the platform was a global collaboration involving researchers from Oxford, Harvard, Northeastern, The Gorgas Institute, Boston Children’s Hospital, Georgetown, the University of Washington, and the Johns Hopkins Center for Health Security.

The OxRSE group has been involved from the outset, contributing to architecture planning and software development alongside the initial team from Google.org. They have also trained students in processing automated data sources and have overseen the ongoing development and expansion of the platform to address outbreaks such as Mpox, Ebola, and Marburg. Currently, four OxRSE staff members are actively working on the Global.health platform, engaging in a range of projects, including graphical analytical pipeline development, the creation of clinical and multimodal data pipelines that incorporate meteorological and mobility data, and the application of large language models for structured data extraction and epidemiological use cases. OxRSE staff have made significant contributions across all aspects of the platform, from frontend development in TypeScript to backend development, data pipelines with Python, and infrastructure provisioning using MongoDB, AWS services, Kubernetes, and Terraform.

GRAPEVNE is a desktop application that offers a graphical environment for building and validating hierarchical analytic workflows, utilizing the Snakemake workflow manager. Initially designed to address phylogenetic challenges, GRAPEVNE has evolved to support a wide array of use cases, including ETL (Extract, Transform, Load), mathematical modeling, and distributed analysis.

InsightBoard is an open-source, flexible, and locally deployable digital tool designed to address the challenges of data integration, error identification and cleaning, harmonization, and visualization during infectious disease outbreaks. The tool integrates Large Language Models (LLMs) for faster data standardization and has been successfully tested during the ongoing 2024 Mpox public health emergency in collaboration with Africa CDC.

Clarabel.rs is a Rust-based implementation of an interior point numerical solver for convex optimization problems, featuring a novel homogeneous embedding approach. Widely adopted, Clarabel has seen nearly 1 million downloads per month for its Python wrapper. The OxRSE team is enhancing Clarabel.rs by incorporating GPU acceleration, significantly boosting solution speed.

PETRUSHKA is a tool designed to facilitate real-time interaction between medical clinicians and their patients. It assists in guiding both the patient and clinician in selecting the most effective and suitable antidepressant based on the patient’s specific medical characteristics and preferences regarding potential adverse events. The tool is currently undergoing clinical trials in the UK and Brazil.

Galv is a battery data management platform designed to enable battery scientists to input and retrieve data along with its associated metadata. The platform ensures that data is clean, robust, and accurately linked to the corresponding cells and cycler machines, thereby enhancing interoperability and enabling meta-analysis. Its code is open source and available on