Data Centre app and model publishing platform is rapidly growing
SciLifeLab Serve is growing in content and attracting recognition as a key platform for publishing and sharing advanced machine learning models and apps – data science application services. To date, over 100 apps and models have been published on the platform, supporting Swedish data-driven life science. Recent developments include offering the possibility for tracking machine learning workflows, regardless of where training and analysis is carried out.
SciLifeLab is expanding its e-infrastructure to support AI-driven life science, with SciLifeLab Serve playing a key role in this transition. Since its beta launch in 2024, Serve has offered the Swedish life science community streamlined access to machine learning (ML) model serving, app hosting, and web-based development tools. The service is free for researchers at Swedish universities and institutions. The platform is also used by SciLifeLab itself to host tools and applications from the infrastructure units.
“To fully realize the potential of AI in life sciences, it is essential that models are not only developed but also made easily accessible to researchers through robust hosting and serving infrastructure. At SciLifeLab, providing scalable and open access to AI models and apps is a key part of our strategy to empower the scientific community and accelerate data-driven discoveries.”, says Ola Spjuth, AI Lead at SciLifeLab.
SciLifeLab Serve is a key component in SciLifeLab Data Centre’s growing ecosystem of services that facilitate Open Science and FAIR sharing of research outputs, complementing data repositories and open data portals. An important upcoming feature is persistent identifiers, such as DOI:s (Digital Object Identifiers), associated with each published item. Serve also provides the ability to keep items private during ongoing development and peer review.
The Serve platform strives to provide a suite of state-of-the-art tools for users to build and use ML models, and to empower the life science community to share their own developed services on a scalable production e-infrastructure environment. While most hosted apps are custom developed apps, Serve also supports apps built using for example Shiny, Plotly Dash, TissUUmaps, Streamlit, and Gradio.
”To date, the majority of SciLifeLab Serve´s users, both from infrastructure and the research community, are based in Sweden, with representation from all major universities. The platform offers advanced tools for publishing and sharing ML models, along with customised virtual, or onsite support from our team of engineers”, says Johan Alfredéen, AI data engineer at SciLifeLab Data Centre.
SciLifeLab Serve has attracted attention from all over the world, to-date there are views from 87 countries.
Moving forward, new features are continuously added to the platform. Most recently, support for TissUUmaps apps was introduced- a tool that enables fast visualization and exploration of millions of data points across tissue samples. In addition, SciLifeLab Serve now allows the possibility for users to manage machine learning models, including integrated experiment tracking via the open-source suite MLFlow.
New and old users are encouraged to contact the SciLifeLab Data Centre at serve@scilifelab.se for support with onboarding models and apps to the platform. The team is continuously developing new features, and user feedback will play a key role in driving continuous improvements.
What are our users saying:
”SciLifeLab Serve engineers put a lot of effort into deploying my models so I could focus on developing our models without having to worry about infrastructure, and deployment challenges.
The service’s reliability and scalability made it the perfect solution for my needs”
Srijit Seal, MIT
”The SciLifeLab Serve was the perfect space to host our interactive heatmap app […]reliable and easily accessible way to share our complex data with our users. The SciLifeLab Data Centre Serve team was incredibly helpful during the entire process […] and providing fast support whenever needed”
Kalliroi Sdougkou, and Stefano Papazian, Stockholm University
”We used SciLifeLab Serve for hosting practicals during the SciLifeLab Omics Integration and Systems Biology course, and it significantly streamlined the process. Ensuring that all attendees had the correct computing environment for each practical was a challenge in previous editions of the course”.
Rasool Saghaleyni, NBIS
