Protocol Browser for Stem Cell Differentiation
The project
The Protocol Browser for Stem Cell Differentiation allows researchers to quickly and conveniently compare and explore stem cell differentiation protocols (methods to develop differentiated cells from pluripotent stem cells in lab environments). The product solves a major inconvenience for researchers, as stem cell differentiation protocols are currently scattered across many sources and are not standardized, making them difficult to explore and compare.
The browser allows users to sign up and submit their own protocols. They can choose to enter protocols manually or use AI-assisted processing. Moreover, anyone can access this browser and query all public protocols. The query feature is aided further by convenient visualisation methods, multi-protocol visualised comparisons and the ability to view in more detail. Additionally, the website features a Lineage Tree page that visualises relationships between different cell types. Through this page, users can explore protocols leading to selected cell types, compare Growth Factors and Stage Markers, inspect protocol connections, and analyse Growth Factor usage across differentiation pathways.
To maintain data quality, the platform includes an administrative system that allows administrators to approve users, manage protocols, and oversee platform operations.
The platform thus greatly improves accessibility of protocol information and helps these researchers make better-informed decisions. By centralising available data through a collaborative project and providing intuitive visualisations it supports more efficient research in stem cell biology and regenerative medicine.
The customer
Our client was Micha Drukker from the Leiden Academic Centre for Drug Research (LACDR), whose research focuses on stem cell biology and regenerative medicine. However, our primary contact throughout the project was with Rafaella Buzatu, a PhD candidate at LACDR.
We used a collaborative and two-directional communication process, holding meetings every 1-2 weeks to present our changes and listen to our client’s feedback and requirements. R. Buzatu, along with Tiago Scarano, a student who worked on the protocol part of the project, provided in-depth feedback and information on stem cell differentiation and the needs of researchers working in the field. Through this frequent, consistent and constructive contact we were able to provide a product that successfully met the client’s requirements and expectations.
The team
Rapolas Bernotas was our Product Owner, with Maria Politylo taking the role of Scrum Master. Cris van der Lugt gradually took responsibility for repository management and Git support. Initially, we divided the project into three separate parts: the user system, the visualisations and the query page. We divided the six team members (the three above, Dylan Moscoviter, Ghazaleh Nasrollahi and Bram Westerbeek) evenly over these three parts. Gradually, as the core functionality was completed and new, smaller issues emerged, we deviated from this approach, adopting a more team-wide angle where we would divide individual features, bugs, etc. appropriately among the team.
While our agile and configurable approach worked very well in some areas, some parts fell behind. It took some time to establish the most effective workflow. As the project evolved, our workflow became less dependent on the original board structure and more focused on direct coordination around smaller integration tasks and bug fixes.
However, we are proud of our efficient progress, as we achieved our Minimum Viable Product (MVP) mid-May, leaving us plenty of room to complete more of the client’s wishes and create a product we are proud of without the pressure of a last-minute development sprint.
The technologies
Languages:
- TypeScript, HTML, CSS for frontend development
- Python for the backend
Important Frameworks, Libraries and APIs:
- React for our frontend user interface
- Vite for frontend build tooling
- Flask for our backend framework
- Firebase Hosting and Firestore for hosting and database infrastructure
- OpenAI API for protocol name normalization and AI-assisted protocol processing
Tooling:
- GitHub Actions for continuous integration and deployment
- Python unittest module for backend unit testing
Deployment:
- The website is hosted on Firebase Hosting and deployed automatically through a CI/CD pipeline connected to the client’s GitHub repository. Changes merged into the main branch are automatically deployed to the live environment.