ADR-24-01: PSDC Pangolin
Status |
Accepted |
Impact |
Low |
Demand Owner |
Tatyana Doktorova |
ADR Contributors |
Joerg Schmiedle, Matthieu Croissant, Padmanabha Udupa, Orlando Pereira, Alex Weigel |
Reviewers |
Architecture Council |
Informed |
Solution Architecture CC, pRED Architecture CC |
Publication Date |
June 6th, 2024 |
đ Background
Pharmaceutical Sciences Digital Products (PSDC) aims to build a comprehensive database, Pangolin (Pan-global data integration), of historical external and internal in vivo and mechanistic in vivo animal and in vitro data generated both externally and internally, with unified terminology. This data should ideally be connected to clinical data (e.g., DCH). The database should be extensible, i.e., able to be linked with other data relevant to computational toxicology. The database and its further development will increase the usability of these data to support drug discovery projects.
âšī¸ Context
Tatyana Doktorova, Dragomir Ivanov Draganov, and Laura Badi from the PSDC presented Pangolin to the Architecture Review Forum and sought guidance on the technical and architectural design of the solution. Discussions surrounding the business architecture, information architecture, information flow, solution architecture, and infrastructure were conducted.
đ Consequences
Failure to adhere to the blueprints and recommendations of the Architecture Playbook will result in the creation of a computerized system that:
-
Introduces fragmentation into the landscape.
-
Increases the probability of future risks related to maintainability and scalability.
-
Deviates from our established guidelines and best practices.
Ensuring compliance with these architectural guidelines is crucial for building a robust, scalable, and maintainable system that meets our long-term objectives.
đ¤ Outcome / Decision
Following the pRED Architecture Review Forum on March 27th, 2024, the Proof of Concept (PoC) for Pangolin has been endorsed. However, several actions and considerations were highlighted during the forum to ensure compliance and optimal progression.
Actions
-
System Risk Assessment (SRA) and Data Classification Report (DCR): These assessments are mandatory to ensure compliance with data security and privacy regulations. Given that the PoC will involve authentication and Roche data loading, the data classification will be C3 and/or 4.
-
Future Inspection: The forum intends to inspect the PoC before moving towards a production system. Key inspection points include:
-
Persistence of data
-
Integration of data
-
Data access control
-
Recommendations
-
Data Engineering Role: It is recommended to involve a Data Engineering role in the daily activities of the project. This will significantly enhance the project’s data handling and integration capabilities.
-
Guidance from the [Data & Workflow Engineering Capability Circle](https://sites.google.com/roche.com/dataworkflowengineering/home): Reach out to this capability circle for guidance on data engineering best practices.
-
AirFlow for Workflow Management: For future stages beyond the PoC, consider the use of AirFlow instead of relying solely on current Python scripts.
-
Compound Search Evaluation:
-
Evaluate the volume of molecules to be searched and select technology accordingly.
-
Prioritize the reuse of existing solutions where possible. Multiple existing solutions can provide similar functionality, such as the compound platform Biovia Direct Cartridge.
-
More documentation is available [here](https://pred-arch.pages.roche.com/playbook/pred-arch-playbook/100-knowledge-base/chemical-structure-search.html).
-