Discovering the Power of teal: From Pre-built Modules to Custom Creations
Welcome to the world of teal, an open-source R package designed specifically for exploratory data analysis and visualization in clinical trials. Whether you’re a seasoned R programmer or just starting your journey, this workshop led by Dony Unardi, Principal Data Scientist at Genentech, offers a comprehensive guide to understanding and utilizing teal for interactive and reproducible data analysis.
A Journey Through teal
In this session, Dony Unardi takes us from the basics of creating a simple teal application to the complexities of custom module creation, all within a three-hour workshop. Let’s dive into the key takeaways and insights from this engaging session.
Introduction to teal
teal is built on the Shiny framework, utilizing Shiny modules to offer a modular and standardized approach to building applications. This modularity allows for flexibility, collaboration, and crowdsourcing, enabling users to create robust applications with pre-built and custom modules.
The session begins with a simple agenda: understanding teal, diving into coding, and exploring additional resources. The goal is to familiarize participants with the teal framework and empower them to explore and create their own teal applications.
Setting the Stage
Before diving into the code, it’s essential to understand the core components of a teal app: data and modules. The init function is the cornerstone of any Teal app, requiring a data argument (a teal data object) and a modules argument (a list of teal modules).
Creating a teal Data Object
A teal data object is created using the tdata function, which requires name-value pairs representing the datasets you wish to include. For example:
data <- tdata(x1 = iris, x2 = mtcars)This object not only stores the datasets but also captures the code needed to recreate them, ensuring reproducibility.
Building Modules
Modules in teal can be pre-built from existing packages like teal.modules.general or custom-built to meet specific needs. The session covers the basics of creating a custom module, starting with Shiny’s UI and server structure and integrating teal’s unique features.
Hands-On Workshop
The heart of the session is a hands-on workshop where participants build a teal app from scratch. Dony guides us through creating a “Hello World” module, then progressively adds complexity by incorporating data selection, visualization, and interactivity.
Creating a Custom Module
The process begins with setting up a Shiny module structure, using the module function to encapsulate the UI and server logic. Key features include:
- Namespace Handling: Ensuring UI components are uniquely identified.
- Reactivity: Using Shiny’s reactive programming to update UI elements based on data inputs.
Participants learn to create dynamic dropdowns for dataset and variable selection and to render outputs like tables and plots using packages like DT for tables and ggplot2 for plots.
Exploring Advanced Features
As the workshop progresses, Dony introduces more advanced features like:
- Predefined Filters: Using
teal.slicesto set default filters for datasets. - Reproducibility: Implementing a “Show R Code” button that extracts the code used to generate the current state of the app, enhancing reproducibility.
- Customization and Extension: Leveraging
teal’s modular structure to extend functionalities and create tailored solutions.
Conclusion and Beyond
By the end of the workshop, participants have a solid understanding of both pre-built and custom module development in teal. This knowledge opens the door to creating sophisticated, interactive applications for clinical trial data analysis and beyond.