Over 70 tranSMART developers and stakeholders met at Vrije University in Amsterdam, from June 17th-19th, for the second tranSMART Foundation Developer’s Hackathon. It was coordinated and chaired by Kees Van Bochove (The Hyve, TraIT), Ioannis Pandis (Imperial College, eTRIKS) and Terry Weymouth (U. of Michigan). A wide breadth of attendees was present with representatives from 11 Universities, 10 life sciences companies, a standards organization (CDISC) and (many) Commercial Service Providers.
The packed agenda included a developer’s track and a business track. The developer’s track focused on the design of a revised tranSMART core, including specification of the target tranSMART architecture. The business track consolidated feature requests and began development of a consolidated tranSMART product roadmap.
Its all about outcomes
Professor Gerrit Meijer (VUMC, Lead of the TraIT Consortium) launched the meeting, welcoming the participants and providing the vision of translational data management and its impact to the lives of patients. He positioned tranSMART as a key driver for improving patient outcomes through its ability to integrate various streams of research. It can serve as the ‘engine’ for translating research to the bedside.
Burgeoning success
tranSMART Foundation Co-Lead Brian Athey (U. of Michigan) and CTO Yike Guo (Imperial College) led a “State of the Foundation” discussion. The tranSMART Foundation is based upon a bold premise:
“What if…. Scientists work at every university, non-profit research organization, pharmaceutical and biotechnology company in the world could share their pre-competitive and molecular biology data – working together to turn scientific discoveries into medical breakthroughs?”
To achieve this, value generation and sustainability are of primary importance. Architecture development, release and content management are key deliverables upon which business models are to be built.
It was highlighted that the workshop is meant to celebrate achievements to date, consolidate the community, structure the process going forward. Most important is ensuring the timely release of tranSMART v1.1 in September.
tranSMART v1.1 introduces the PostgreSQL open source database base management system (DBMS). The PostgreSQL integration is the result of a combined effort between the University of Michigan and Imperial College (eTRIKS). It is critical for continued tranSMART adoption by academic institutions as well as public-private translational research partnerships such as the TraIT (Netherlands) and eTRIKS (Innovative Medicines Initiative [IMI]). As the Oracle database engine is likely to remain the preferred choice by many commercial organizations, plans were made to qualify both DBMS products with tranSMART v. 1.1.
With the increasing number of developers there is a pressing need for structure. The recommendation is that special interest, or working groups be established.
The way to the future
Project roadmaps for TraIT, eTRIKS, EMIF (IMI) and the tranSMART Foundation were presented.
TraIT: a study driven approach that includes an improved user interface, security, extension of data exploration functionalities and high dimensional data
eTRIKS: translational research study repository, an open analytics platform, translational research findings and knowledge repository, a collaborative environment, and scientific evidence and provenance support.
EMIF: process for privacy protection and user authentication and authorization, across trial query, data harmonization.
tranSMART: movement towards a framework architecture, ability to support 1000’s of studies, dynamic module loading.
In order for the developers to better understand the needs of client roles, user patterns were reinforced by example scenarios with tranSMART clients, or “avatars”, representing the needs of Bioinformaticians, Clinicians, Epidemiologists, Pathologists and Molecular Biologists.
The participants attended sessions of their choosing during most of days two and three. Ioannis Pandis chaired the business track where Emmanuel Van Der Stuyft (Johnson & Johnson) established a framework for eliciting and aggregating business requirements across stakeholders.
Many of the top requirements were shared across projects including:
- support for longitudinal studies
- data provenance
- cross study comparison
- NGS, imaging, animal models
- user interface/visualization extensions.
A requirements working group was chartered to continue to refine the business requirements into a deep product backlog of valued enhancements.
‘Build it and they will come’
The technical track, chaired by Kees Van Bochove and Terry Weymouth, specified a first draft of the comprehensive target tranSMART architecture. Development activities will be devoted to core:
- API (function definitions)
- application (basic services and plugin registries)
- distribution (reference installation)
Each core element will have a corresponding working group. The developers also settled on the release process for version 1.1, including recurring teleconferences, test planning and JIRA bug tracking to best ensure version 1.1’s timely delivery.
tranSMART community is a vibrant, committed and growing ecosystem welcoming of academics, commercial users and commercial service providers. It is becoming more and more of a thriving open source community. This is demonstrated by the broad attendance and participant enthusiasm and even more so by the tangible outcomes produced:
- The establishment of the tranSMART Foundation as the center of community coordination.
- Momentum maintained by a shared focus on a successful and timely tranSMART 1.1 release.
- Stakeholders that are aligned with regards to developing tranSMART and target architecture and business requirements that emerge easily by consensus.
- Clear evidence that a rare universal information platform that helps drive data-intensive translational research is in the making.
The next developer’s workshop is planned for October to follow the tranSMART version 1.1 release and for planning the next set of milestones.
- Tweet
-