Trustworthy and Adaptable Biological Data Integration

Overview

In modern science, the key to insight often lies in merging complementary views. This is perhaps nowhere more evident than in biology and medicine, where the relevant molecular processes are only indirectly observable, yet decreasing costs mean many views are cheaply attainable. This has presented rich opportunities for statisticians who can navigate diverse data and form coherent scientific narratives from them. But the techniques that drove the first wave of progress are beginning to hit their limits: methods that impose rigid constraints on the data or design are no longer suitable, modern workflows complicate reproducibility and interpretability, and community resources are not effectively utilized.

This workshop aims to spark creative and productive exchange on these challenges. We will spotlight recent efforts that extend high-dimensional statistics and statistical software to current problems, from generative models that align isolated experimental results onto community-level maps to data-driven simulators that allow systematic calibration and benchmarking. We envision statistical technology that enables practitioners to navigate the emerging scientific landscape and apply adaptable, trustworthy methods to diverse data integration problems. Progress depends on effective communication across statistics, biology, and machine learning, and our workshop will bridge these communities — helping biologists learn novel data analysis principles and guiding computational researchers toward the abstractions most needed in practice.