SciLifeLab Serve opens for users to share machine learning models and data science apps

SciLifeLab Serve (https://serve.scilifelab.se) is a service from the SciLifeLab Data Centre for sharing machine learning models as well as data science applications, such as Shiny and Plotly Dash. It is now open for beta testing to all users.

The development of new cutting-edge services that enable data-driven life science research is an important part of the SciLifeLab and Wallenberg National Program for Data-Driven Life Science (DDLS). SciLifeLab Serve is intended for use by individual researchers, research groups, and SciLifeLab infrastructure units. The service allows users to e.g. share trained machine learning (ML) models with the scientific community or host prototype services (Gradio, Streamlit, FastApi, Flask etc.) built on top of a trained ML model as part of the commercialization efforts. In addition, it allows users to host data science applications (Shiny, Dash, etc.).

SciLifeLab Serve is an important part of the SciLifeLab digital infrastructure Data Centre is building.

The service is free of charge for life science researchers affiliated with a Swedish university or research institution. SciLifeLab infrastructure units that drive innovation and methods development in their respective areas will also be able to use Serve to host developed tools, especially those that use ML technology.

SciLifeLab Serve has already been used by some researchers, and they showcase the breadth and flexibility of the service.

• ”We have set up our database about the association between the intestinal flora and the plasma metabolome, GutsyAtlas, as an app on SciLifeLab Serve. The app allows for users to create their own tables and figures based on which bacteria or metabolite they are interested in. We received fantastic support from the SciLifeLab Data Centre Serve team in the process. Using SciLifeLab Serve had several advantages for us. Compared to the commercial alternative it was, of course, a lower cost. In addition, Serve also offered more space for data, a reliable Swedish URL and faster server. All of which is important when there are such large amounts of data”, says Tove Fall, professor at the Department of Medical Science at Uppsala University. 

• ”SciLifeLab Serve has allowed us to host the methylR app, a complete tool for Illumina DNA methylation array data analysis, and make it easily available for our users. The metylR app is primarily used by researchers. It offers data visualization, normalization as well as additional features such as the annotation of the genomic features packed in a minimal, elegant and intuitive graphical user interface,” says Jyotirmoy Das, Principal Research Engineer, Linköping University and Clinical Genomics Linköping, infrastructure unit at SciLifeLab.

”Over the past year, we have carried out pilot testing of a prototype of SciLifeLab Serve with a group of users that included researchers and technology experts from SciLifeLab infrastructure units. Over 70 different applications and models have already been shared through SciLifeLab Serve. These users’ feedback allowed us to make improvements in functionality and usability before the rollout of this new beta version, now open to all,” says Ola Spjuth, Professor at the Department of Pharmaceutical Biosciences, Uppsala University and Scientific lead of SciLifeLab Serve.

Researchers can receive support from the SciLifeLab Serve team in onboarding their models and apps into the platform, they can send their requests to serve@scilifelab.se.  The team is planning to implement and roll out many new features in the future. User feedback will be crucial for future development of SciLifeLab Serve.