Background

• Challenges with productivity, efficiency, and innovation in the pharmaceutical industry

• Flat number of new drug approvals per year, but increasing R&D spend per approval

• Several potential reasons:

  1. New, high-risk therapeutic areas
  2. Advances in technology leading to a wider range of potential targets
  3. Increased competition and regulation

Delivering R&D value

Methodological innovation is required due to:

• New data modalities and endpoints

• Higher volume of data

• Novel designs

• […]

Why open-source statistical software?

• Speed:
  • Implementation of innovative designs and statistical methods
  • Collaboration with academia and between companies
  • Shared experience with potential bugs and edge cases
  • Rapidly evolving needs

Why open-source statistical software?

• Scale:
  • Sharing the burden of developing, maintaining statistical packages
  • Joint, faster adoption across the industry
  • Developing talent and capability

Why open-source statistical software?

• Transparency:
  • Valuable for audits
  • Valuable for regulators
  • Valuable for further developments

Challenges

• Regulatory and validation hurdles:
  • Documentation, versioning, testing, and traceability requirements for submissions
  • The R Validation Hub is working on developing validated systems fit for submission to FDA/EMA

Challenges

• Organizational and cultural barriers:
  • Proprietary mindset, reluctance to share code due to perceived IP risk
  • Departmental silos, creating communication gaps between departments (e.g., IT and statistics)
  • Purpose silos, different software packages for different tasks
  • Project silos, lack of reusable/general code shared across projects

Challenges

• Technical quality and usability risks:
  • Usability gaps
  • Poor documentation and inconsistent interfaces may limit uptake
  • Cross-language and interoperability issues for multi-language teams

Challenges

• Software quality:
  • Lack of testing
  • Lack of reproducible workflows
  • Increased need for professional research software engineers

Challenges

• Sustainability:
  • Maintenance risk, single-author projects can stagnate without funding or contributors
  • License complexity (in a legal sense)

The way forward

• Risk-based validation: align package risk level with validation effort and documentation

• Usability first: API design, examples, and user testing as core deliverables

• Break silos: cross-functional governance and shared infrastructure teams

The way forward

• Create RSE roles: embed software engineering expertise in biostatistics groups

• Develop engineering standards: version control, CI, automated tests, release management

• Leverage the community: openstatsware, pharmaverse, and other cross-company initiatives

Key takeaways

1. The open-source revolution in the pharmaceutical industry is in full swing

2. This is exemplified by successful submissions by Novo Nordisk and Roche (among others)

3. Open-source software offers faster innovation, transparency, and shared efficiency if paired with governance, validation, and good software engineering practices

Minimum set of good practices

• Documentation

• Vignettes

• Tests

• Functions

• Style

• Life cycle

Developers Value Tests For Software Longevity

Documentation

Documentation is important for users and developers to understand all objects in your package, without reading and interpreting the underlying source code.

• Use inline comments next to functions, classes and other objects to generate their corresponding documentation

• Do document internal functions and classes for maintenance by future developers

• Add code comments for ambiguous or complex pieces of internal code

Vignettes

Vignettes are documents that complement the documentation by providing a comprehensive and long-form overview of the package from a user perspective.

• Provide an introduction vignette that introduces the package to new users

• Include code examples and automatically compile the vignette to ensure reproducibility

• Include deep dive vignettes that go into depth on specific use cases, functionalities or underlying theory

• Host your vignettes on a dedicated website

Tests

Tests are a fundamental safety net and development tool to ensure that your package works as expected, both during development as well as on user systems.

• Write unit tests for all functions and classes in your package, to ensure that all building blocks work correctly on their own

• Write functional tests for all user-facing functionality, to ensure that the package is stable when refactoring internal code

• Ensure adequate coverage of your code

Functions

Function definitions should be short, simple and enforce argument types with assertions.

• Write short functions for a single and well-defined purpose, with few arguments, and low complexity

• Use type hints to explain to users which argument of the function expects which type of input

• Enforce types and other expected properties of function arguments with assertions

• Catch errors and fail early with human-readable error messages

Style

Code style is important because it makes software easier to read, safer to change, and more consistent for everyone who works on it.

• Use idiomatic code and follow clean code rules

• Use a formatting tool to automatically implement a consistent and readable code format

• Use style checking tools to enforce a consistent and readable code style

Life cycle

Life cycle management is simplified by reducing dependencies, and should include a central code repository.

• Reduce dependencies to simplify maintenance

• Only depend on other packages that you trust

• Give clear information on user-facing changes in the package, and first deprecate functionality before removing it

• Use a central repository for version control, collecting and resolving issues, and managing releases

Resources

• openstatsware website:
  https://www.openstatsware.org

• Current version of our openstatsguide:
  https://www.openstatsware.org/guide

• Good Software Engineering Practice for R Packages course material:
  https://openstatsware.github.io/shortcourse-iscb2025/

Check our website: reddooranalytics.se/career