The pharmaverse is a connected network of industry individuals working together in collaboration to achieve the below objective. We are not a consortium nor organisation, and we have no binding links to any other industry group. We receive no funding and all contributors do so of their own accords and represent their own ideas, not those of any single company. Any legal counsel comes from legal representatives from any organisations of our council members.
The objective of pharmaverse is to bring together a collection of pharmaceutical industry representatives to agree on a curated and opinionated subset of open-source software packages and codebases, which would be based on the R language and those that we believe are the most suitable for the industry to adopt in order to deliver the clinical data pipeline. If suitable tools do not already exist within the open-source ecosystem, these gaps can be identified. Then via collective collaboration (around development and testing), new solutions may be developed with the aim to reduce duplication of efforts and gain increased harmonization to the way we work across the industry.
There would be no commercial interest from contributors to any of these efforts, as all code will be targeted for open-source release under a permissive license.
The benefits would be as follows:
- Industry collaboration leading to more robust and well-thought-out solutions with shared development and maintenance efforts, thus distributing the cost and burden across multiple organizations
- Greater potential to bring unified solutions to regulators demonstrating collaboration between industry and standards organization to form workable solutions
- Better enablement of our medicines to reach patients and society faster through the pooling of skills and talents and accelerating how the industry generates insights from patients’ data - and to do this in a more sustainable way
- Attraction of the next generation of great software developers and data scientists to Pharma and providing increased transparency of our industry
The initial focus of pharmaverse is on the journey from CRF through to submission for clinical trial analysis reporting via R packages. We intend to provide a pathway to share efforts to build open-source codebases, to identify and rectify gaps in available tools, and then collaboratively build and maintain any additional codebases that complete this end-to-end workflow. This could include, for example, solutions covering SDTM mapping, ADaM derivations, TLG production, and enabling eSub for both today’s expectations and potentially through interactive tools in the future. It also could include utility packages that may help this dataflow in any way, such as diffdf, which performs data frame comparisons.
The R packages recommended under pharmaverse will fall under one of the following:
|External to pharma||Transcends specifically pharma needs||tidyverse, diffdf|
|Pharma-specific independent of pharmaverse||Created for use in pharma, but not necessarily following the pharmaverse charter and recommendations||rtables, xportr|
|Pharma-specific under pharmaverse||Created for use in pharma, following the pharmaverse charter and recommendations||admiral|
The aim will not be to agree on cross-industry implementations of CDISC standards, as each company may have their own repositories of chosen implementation rules and specifications. Any open-source package or codebase would knowingly act as a starting point for code re-use, yet will never be able to cover all eventualities. By following this principle, we would allow the framework to take on a level of standard agnosticism, which improves the likelihood of remaining future-proof.
The pharmaverse environment should be easily locatable and accessible via a single site. Some packages will already exist on CRAN or GitHub, whereas others are either in ongoing development or brand-new packages that will come from future pharmaverse collaborations. For the already open-source released packages, we could collate and explain clear use cases for clinical trial reporting. Some may cover similar purposes (e.g. there are numerous offerings for TLGs) so we should attempt to share the differences and unique merits to help a company or user choose which packages or tools they might adopt.
There are different levels of commitment companies and individuals can make towards pharmaverse, as detailed below:
|Council representative||Strategy and decision-making body across pharmaverse|
|Working group member||Support selection of pharmaverse preferred packages in dedicated SME areas|
|Package co-lead||Lead and maintainer for a specific pharmaverse package or codebase|
|Individual contributor||Development contributor to any pharmaverse package or codebase|
The added benefits of contributing to pharmaverse range from an individual having the opportunity to share their code wider and make new connections across the industry, through to a company getting early awareness and access to prototyping and influencing future solutions.
Contributing to pharmaverse is open to all organizations working within the clinical trial analysis pipeline, creating relevant code. An inclusive model will be used to help identify gaps and, from this endorsement, active participation on individual codebase gaps (like the admiral project) are on an opt-in basis, i.e. just because a company contributes to certain pharmaverse projects it doesn’t mean you have to contribute to all.
The pharmaverse recommendation is for all our new code collaborations to be hosted under github.com/pharmaverse. This forms a neutral home for all open-source repos falling under our scope. The admins of this org would include pharmaverse council members.
The preferred model of development of new pharmaverse offerings would be to initiate projects as open-source from the start of development or as soon as a stable first draft is available. This is coupled with an expectation of engagement of other companies in ongoing testing and possibly via supporting development efforts with pull requests/issues/feedback. This way we result in more robust code for which many companies have already engaged with during the development process, and it has a higher chance of successful embedding. In addition, the open and transparent development reduces the chance of others creating duplicative parallel packages.
For more information on development guidance and requirements, see the minimum inclusion criteria and recommended practices available via our Package co-lead guidance.
Risks & Testing
Each respective company should decide whether legal contracts or collaboration agreements are required. Any company involved should be free to decide to withdraw from any pharmaverse collaboration at any stage.
Testing of any code has to be a key consideration. Packages initiated to fill gaps in the pharmaverse will need to meet the required standards as outlined in our Package co-lead guidance. It will also be necessary to make available any evidence generated on the reliability and methods executed, as this transparency would in turn encourage wider engagement and adoption of the package.
Liability for the reliability of the pharmaverse environment remains with the users, but to help limit risk, package dependencies should be intentionally limited to the most necessary, such as selected tidyverse packages.
For any co-development effort, there should be no need to access any other company’s systems and no individual company patient-level data should be shared.
Governance & Maintenance
The pharmaverse recommendation is for the control of the main branch of any package repo to be limited to a small number of “Package co-leads”, preferably from different companies, that have a shared need to invest in a common solution. Each respective development team is ultimately responsible for planning the ongoing maintenance strategy, but representatives to support this could come in from other companies, such as those involved in testing. Feedback or pull requests could come from any “Individual contributor”.
If you imagine the case of a codebase (as opposed to a single package), then a similar model to Wikipedia could be imagined, where numerous representatives from across companies each take responsibility to “garden” a domain and ensure any user contributions are appropriate to be accepted into the codebase.
Each development team will agree a cadence for their own releases and specific targeted communications. The council should then supplement this with an aligned method of regularly communicating iterations of the pharmaverse environment out to the industry, as well as the wider community of Open Pharma aligned projects.
Why is now the right time for this? The industry has never been more conducive to cross-company collaboration (considering efforts like PHUSE, TransCelerate, Accumulus Synergy, and more). Nowhere was this witnessed more than in how the industry and regulators partnered to develop guidance with urgency around how to manage the impact of the COVID-19 pandemic. Furthermore, in the field of clinical reporting, we stand at a possible unique point in time where many companies are starting to embrace alternatives to what has been their programming language of choice for 30+ years, by integrating open-source languages such as R. At this juncture, we have the opportunity not to all take separate routes on this journey, as may have been done before. The pharmaverse offers a chance to come together collectively as an industry at a magnitude that’s never been seen before in clinical study programming. Our patients are waiting and expecting.