Posters - Schedules

At AstraZeneca, we harness data and technology to transform how we discover and develop new medicines. Hundreds of people across our Research and Development (R&D) organization have roles that involve data and artificial intelligence (AI).
To enable our scientists to leverage data and AI, we need to increase their awareness of what data science can do; similarly, we need to supercharge connections between our data scientists, helping them sharpen the skills they need to impact the pipeline.
Over the last two years, the Data Science & AI’s Learning & Development team has developed a portfolio of learning opportunities including degree programs, short courses, and e-learning and has built a cohesive data science community with a flag-ship data science symposium. Our activities cater to different audiences, ranging from the skeptical, who previously failed engagements with data science projects, to those who see AI as a panacea to operational and research projects.
Here we present the progress we have made on our journey to build an engaged Data Science learning community with thousands of touchpoints across AstraZeneca R&D, discuss the challenges that we face and present some of our success stories, from technical training to improving the way we communicate across expertise.

Identifying transcriptomic changes across experimental groups via bulk RNA-seq has become a common approach in biological research. DESeq2 is one widely used package that addresses key challenges in differential expression analysis. However, DESeq2’s approach can be difficult to understand depending on its users’ statistical backgrounds, and particularly in cases where biologists want to better understand analyses conducted by their bioinformatics collaborators. Here, we present a teaching tool intended to familiarize users with the intricacies of DESeq2’s approach. Unlike existing vignettes and tutorials, this tool is interactive, allowing users to explore parameters that influence downstream differential expression calls. For example, while learning about dispersion shrinkage, users can change the number of replicates in each sample group using a clickable slider. Their changes are reflected in a responsive plot of shrunken dispersion estimates. After iterating through different sample size values, the user can draw the connection between degrees of freedom and the magnitude of dispersion shrinkage. Interactive learning allows users to drive their learning process and explore nuanced behaviors of DESeq2. As a result, we expect that this tool will improve users’ confidence in differential expression results produced by DESeq2 and will be more comfortable employing it in their own research.

To improve training of life sciences researchers in Brazil, we conceived the Brazilian Python Workshop for Biological Data in 2017, organized by undergraduate and graduate students, and supported by researchers in computational biology, data science, and life sciences. We previously described our workshop, based on the 2020 edition, which started to be organized in a virtual format due to the COVID-19 pandemic. Since then, we have increased our activity on Instagram (mainly) and on Twitter, extending our impact beyond the course week and workshop attendees. Since 2021, we have implemented a short event inspired by the Train the Trainers from EMBL-EBI and Australian universities, that aims to encourage the exchange of experiences between previous and the current organization committee. We also included a lecture with focus on computational reproducibility and programming good practices. This year, we have implemented a code club for the organizers, to help them become familiar with Python and teaching practices. Additionally, we plan to allow donations from attendees, which we expect will benefit organizations with focus on inclusion, diversity, and programming in science. Finally, we are working with the Human Research Ethics Committee to do qualitative research during and after the workshop.

The increasing generation and availability of -omics data, as well as its complex analysis, has created a challenge to biomedical researchers, especially those without formal bioinformatics training. The Bioinformatics Support Program at Yale’s medical library teaches researchers to use bioinformatics tools and perform their own -omics analysis. The program has also evolved to offer professional development opportunities for researchers and to introduce an alternative career option for scientists.

To keep pace with evolving bioinformatics tools and methods, the library created a peer-to-peer (P2P) bioinformatics teaching series. The P2P program helps graduate students and postdocs acquire transferable teaching, presentation, and classroom management skills, while providing unique training to our bioinformatics community and highlighting the library as a collaborative partner.

Another response to the increasing demand for bioinformatics consultations and collaborations was the creation of an 18-month post-doctoral fellowship. This position provided a scientist a chance to explore a career in medical librarianship through professional development, training, and opportunities to present, publish, and collaborate. The fellowship was such a success that a permanent position was created.

The successful implementation of both P2P teaching and an innovative fellowship highlighted the medical library as an unexpected and supportive professional niche for PhDs in biomedicine.

EMBL-EBI provides a range of freely accessible online training, including self-paced tutorials, recorded webinars and materials from live courses.

Feedback from users identified that, particularly for those new to bioinformatics, it can be challenging to identify which of the many online training options are most suitable for them. To assist learners, a collection of introductory online training from EMBL-EBI was developed. This self-paced collection includes online tutorials and videos introducing the topics of bioinformatics and data management, as well as EMBL-EBI resources.

To support learners in the completion of the collection, a blended approach to learning was developed. Asynchronously, the collection encourages learners to provide their own input, and learn from others, by giving their own answers to questions such as ‘what is bioinformatics?’. Two synchronous question and answer webinars were then delivered on the topics of ‘genes and gene expression’ and ‘proteins and structures’. These webinars gave learners the opportunity to ask questions to a panel of EMBL-EBI resource experts.

To expand this blended approach, new collections have recently been released on the topics of chemical biology, biocuration, and finding and using publicly available data, with more collections currently in development, and future question and answer webinars planned.

Professions within the life sciences domain are evolving rapidly with the adoption of new methods and technologies, which requires researchers and other professionals to incorporate new skills to stay at the forefront of the latest developments in their discipline.

At EMBL-EBI we have built a tool to support continuous professional development by facilitating the identification of training needs and the access to relevant training resources. The Competency Hub hosts competency frameworks for different groups of professionals, including the ISCB framework for students and professionals in computational biology. It enables the definition of career profiles, which helps identify the abilities required for a specific role, e.g. bioinformatician, and therefore, inform career choices and professional development.

The tool allows the association of training resources with the competencies, so that users can find where to start learning. It includes learning pathways, curated sets of resources to address a specific challenge within a field or community, e.g. how to access high performance computing resources to run simulations.

The Competency Hub has been developed in close consultation with competency experts and by gathering input from target users through an iterative user experience design approach to make sure that it meets their needs.