Guillaume Desachy is dedicated to helping bring the right medicine to the right patient by leveraging the power of biometrics. As Head of Biometrics at Pierre Fabre, he leads a department of 15+ experts in data management, programming and statistics, working collaboratively to drive the success of both clinical trials and real-world evidence studies.
Since his graduation from the French National School of Statistics (ENSAI) in 2011, Guillaume has worked across all stages of drug development across diverse landscapes, including academia, biotech ventures and global pharmaceutical industry leaders.
In this post, Guillaume discusses how he established a community of R users within the leading pharmaceutical company AstraZeneca. Recognising the growing importance of the open source programming language R within AstraZeneca and the Big Pharma industry at large, Guillaume and his team inspired users with regular challenges and initiatives, and even organised a global conference.
The well-connected and active community has boosted the company’s ROI, and has been celebrated as a model for community-building in the pharmaceutical industry. Reflecting on the project’s success, Guillaume offers his insights on how to build an open source community in your own organisation:
In recent years, the pharmaceutical industry has experienced a significant paradigm shift in its adoption of opensource programming languages. Up until recently, each young graduate had to choose between open-source and proprietary programming languages. However, the landscape has evolved, with a majority of pharmaceutical companies now embracing a multilingual programming strategy. The goal is to use the right tool for the right job.
In early 2021, a group of us at a major pharmaceutical company convened to discuss the growing importance of the open-source programming language R within our organisation and the industry at large. R was not a new thing at AstraZeneca, nor was it a fleeting trend. Despite its established presence, we had limited insight into who the R users were within this large organisation, and there were no formal channels for them to connect, share knowledge, or inspire one another. No forums, gatherings, or events existed to facilitate the collaboration and growth of this community.
Federate all collaborators using the open-source programming language R, enhancing cross-team collaboration and knowledge-sharing.
We decided it was time to establish a community of R users within AstraZeneca. Since 2016, the company has utilised an internal social media platform called Workplace, akin to Facebook for companies, as the primary communication tool among groups. It was immediately clear that this platform would serve as the foundation for developing our R user community.
We decided to steal with pride a successful concept from the external world: the data visualisation contest known as TidyTuesday * , which utilises opensource data. However, we didn’t just want to replicate it; we aimed to make it uniquely our own. The core principle would remain the same as TidyTuesday, but with a distinctive twist. Each time we promoted a dataset, we aligned it with our corporate values and current global events. And that’s how #azTidyTuesday was born!
FIGURE 1
Featured entry for the final edition of #azTidyTuesday
Dataset: Billboard Hot 100
Scientific question: Is there anything that can explain that a song is going to make it to the top of the charts? Could it be explained by its genre, its danceability, its energy, its loudness, its speechiness, its acousticness, its instrumentalness, its liveliness, its valence, its tempo, its duration?
Graph credit: Alvaro Mendoza Alcala
* TidyTuesday is a weekly initiative by the global R community for practising data analysis skills. Every Tuesday, a new dataset is posted on GitHub, and participants are invited to explore, visualise, and share their analyses using the R programming language and its packages, especially those from the tidyverse. It’s a great opportunity to learn, practise, and improve your R skills, as well as to see how others approach the same dataset.
#azTidyTuesday was just the beginning and proved to be a very popular way to promote and grow the community. It also underscored the need to develop additional initiatives to meet the diverse needs of our community members.
Shortly thereafter, we launched a monthly initiative showcasing publicly available R packages or functions. This was done through blog posts that showcased both well-known and lesser-known packages and functions. Each blog post included a few lines of code that readers could try out at their own convenience and pace.
It quickly became apparent that we needed to federate this community of R users, and an internal R conference seemed like the perfect way to achieve this. Such an event would send a strong message both internally and externally.
The very first AstraZeneca R Conference, #AZRC2022, was scheduled for February 2022. At the time, our community was growing but still in its early stages, so we were uncertain about the level of interest, the topics that would resonate, and who would be willing to help organise it.
To our amazement, more than 500 collaborators from 22 countries attended the inaugural AZ R Conference! This conference was a fully virtual, halfday conference featuring Max Kuhn as the keynote speaker, over 30 speakers and nearly 100 poster authors!
The conference was structured into three parallel tracks: governance, machine learning, and reproducibility.
500+ participants from 22 countries!
The overwhelming response to the conference sent a clear message: the community of R users within AstraZeneca was getting definite traction.
Rather than resting on our laurels after the success of this internal conference, we recognised the need to improve inclusion and diversity in addressing the community’s needs and learning styles. We decided to evaluate all future initiatives based on two indices: the self-led learning index and the synchronicity index. The self-led learning index differentiates between self-directed learning and structured instruction, while the synchronicity index distinguishes between synchronous and asynchronous initiatives.
For example, #azTidyTuesday and the ‘Function of the Month’ initiatives were asynchronous by nature, allowing community members to engage at their own pace. In contrast, the AstraZeneca R Conference 2022 was highly synchronous: although all talks were recorded for later viewing, real-time interaction with speakers required live attendance.
Following the inaugural internal R Conference, we knew we needed to continue fostering a sense of belonging and strengthening connections across departments.
We then launched ‘R @ AZ 10:1.’ The idea was to showcase an R user within the organisation and ask them 10 questions, further promoting engagement and collaboration within our growing community.
Some of the feedback we received from the community after #AZRC2022 was a desire for more frequent, conference-like events. We heard this loud and clear, but we also recognised that everyone has busy schedules.
This led to the idea of hosting ‘Lunch & LeaRns’. The concept was simple: invite a member of our community to showcase something cool they did with R, all within a 30-minute session. The only rule was that each Lunch & LeaRn would last exactly 30 minutes, including Q&A.
And again, we kept inclusion & diversity in mind by scheduling these sessions at lunchtime, alternating between Europe and the U.S. to accommodate different time zones.
As more collaborators began returning to office sites, we saw an opportunity to further nurture these virtual connections in real life. This inspired the launch of the ‘R Hot Desk’ at the AstraZeneca Gothenburg site.
The principle is straightforward: once a month, experts in the open-source programming language R meet in an open area of the Gothenburg offices and dedicate two hours to answering R-related questions. This initiative allows for face-to-face interaction and real-time problem-solving, further strengthening the community.
The interest in our face-to-face help desk did not meet our expectations. While there was some definite interest, the format may not have been ideal. Therefore, we decided to rejuvenate the concept by transitioning to an online help desk instead of a face-to-face one.
All of this was based on voluntary involvement: there was no hierarchy between Guillaume Desachy (R@AZ Lead) and the members of the Steering Committee. Given the preciousness of everyone’s time, it’s reasonable to question the time spent building such a community. What is the value of this community? What are the key performance indicators (KPIs) for this community? What is the return on investment (ROI) of this community?
At first, these questions can be puzzling and extremely difficult to answer.
When considering the value of this community, we initially looked at the number of members in our internal social media group. However, this was not a true KPI, as it is not a metric we can directly influence. No matter how hard we worked, our initiatives could only nudge collaborators to join the group. Joining a social media group was never the end goal for this community. The true objective was to foster the community and encourage meaningful connections.
One metric that seems to be a more accurate KPI is the number of ongoing initiatives within the community. The Steering Committee has direct influence over this, as it decides which initiatives to start and run on a regular basis. This metric better reflects our efforts to actively engage and support the diverse needs of the community.
This KPI was used to gauge the vitality of this community. Regularly, we assessed additional KPIs for each initiative led by the R@AZ Steering Committee: Is the initiative well-attended? Is it well-received? Is it meeting the needs of the community?
When momentum around an initiative began to ebb, the Steering Committee would pause, reflect, and sometimes decide to retire the initiative. This was the case with #azTidyTuesday, which we decided to retire after 25 editions.
Key performance indicators are one thing, but they still don’t fully address the question of return on investment (ROI). This question was particularly challenging to answer until one day when an Executive Director at AstraZeneca reached out to Guillaume Desachy and one of his teammates with a simple query:
Neither Guillaume nor his teammate had ever used this R package. However, we knew we could leverage the community. We posted a message in the Hot Desk Teams channel, and within 40 minutes, we had identified someone who had experience with the package.
Reflecting on this, we realised that without the existence of this community, it would have likely taken much longer to identify this subject matter expert. This highlighted the ROI of having a wellconnected and active community.
From the very beginning, our objective has been to federate all collaborators using the open-source programming language R, thereby strengthening collaboration across teams and enhancing knowledge-sharing. In less than three years, we have built a thriving open-source community of 1,600+ collaborators within a large pharmaceutical company. If we consider that all collaborators in this community are connected, it means there are a total of 1.4 million * connections within this community. By this measure alone, we have certainly met our objective of uniting all collaborators using this opensource programming language in AstraZeneca.
As decision-makers, we value tangible KPIs and a clear ROI. One achievement the Steering Committee is particularly proud of is the creation of a psychologically safe space. The once nascent Hot Desk where community members were hesitant to ask questions, has evolved into a place where collaborators of all skill levels can quickly find solid answers. This community has also enabled collaborators to take on leadership roles, whether leading an initiative, heading the organising committee for the yearly conference (20+ collaborators), or even leading the community’s Steering Committee (15+ collaborators). For those looking to develop professionally, this community has been a platform for honing skills such as leadership, communication, and networking.
This open-source community has been repeatedly recognised as a model for community-building within a company in the pharmaceutical industry and showcased in various forums:
useR! 2022,
R/ Pharma 2022,
Posit blog post [2022],
Posit MeetUp [2022],
Posit customer story [2023],
Boehringer-Ingelheim’s guest blog post [2023],
The Effective Statistician Podcast [2023],
R-Ladies Paris MeetUp [2024]
Over a three-year period, this once-emerging community evolved into a hub able to attract worldleading experts for exclusive gatherings.
Although the Lead of the R@AZ Steering Committee has now embraced a new career opportunity as Head of Biometrics at Pierre Fabre Laboratories, the R community he leaves behind at AstraZeneca lives on. A new Lead of the R@AZ Steering Committee has since taken over, and the community continues to grow, meet the needs of its members, and challenge itself. This ongoing effort consistently promotes the use of open-source within a large pharmaceutical company.
A true advocate for the importance of building communities and strong networks to enhance collaboration and knowledge-sharing across teams, Guillaume Desachy leverages his experience with R@AZ to foster collaboration across different departments at Pierre Fabre Laboratories. Striving to be as inclusive as possible, he remains mindful of diverse learning styles in all his initiatives and day-today work as Department Head.
Whether your organisation is large, medium or small, building strong communities in a work environment is crucial. In this article, we have discussed the clear return on investment of such ventures. You too can start your own community of subject matter experts within your organisation. Here are five tips and tricks we’ve learned along the way:
SPECIAL THANKS TO ALL WHO CONTRIBUTED TO BUILDING THIS COMMUNITY, INCLUDING
Abhijit Das Gupta, Anna Samsel, Anna Strzeszewska-Potyrała, Antal Martinecz, Bartosz Górnikiewicz, Chuyu Deng, Daniel Schiff, Ewa Klimkowska, Gabriella Rustici, Gustav Söderström, Jamie MacKay, Jasia King, Jasmine Sadler, Kartik Patel, Lindsey Jung, Monika Eliasz-Kowalska, Monika Huhn, Parth Shah, Per-Arne Stahl, Rachel Evans, Sara Nozohouri, Theodosia Salika, Tom Marlow, Tomasz Lebitko, Vera Hazelwood, Yerka Larsson, and Zach Dorman.
For further insights into Guillaume’s career and publications, you can follow him on LinkedIn
www.linkedin.com/in/guillaume-desachy.
