Data Science Catalyst Program

Key points of the discussion from the Information Session

Project Proposals:

• Applicants are encouraged to submit their project proposals even if they’re uncertain about the level of interest from a data science standpoint. Feedback provided by the Adjudication Committee will assist them in refining their proposals for subsequent rounds in the event of unsuccessful outcome in the first round.
• Typically, proposed projects will likely fall into one of three categories: (1) issues unrelated to data science, but more aligned with software or other disciplines; (2) relatively straightforward problems suitable for a resolution with the assistance of a co-op student; (3) complex data science challenges that align with the DSI’s objectives.
• No specific preference will be given to multiple-PI projects over those with a single PI. The final ranking will be determined by the Adjudication Committee.
• Because of time limitations, the Adjudication Committee cannot offer feedback to PIs before project proposals are submitted. Please submit your project proposal and feedback will be provided.

Support from DSI:

• DSI resident data scientists possess postgraduate qualifications with extensive professional data science experience spanning many years. They are not recent graduates with limited experience.
• DSI Director, Dr. Raymond Ng, will supervise the projects, offering guidance and mentorship throughout the one-year duration of the Program.

IP & External Grant Applications:

• No formal agreement exists regarding the IP potentially arising from the Program. All IP rights are vested in the PI and their team, not the DSI, with no claims to IP by the DSI. There may be exceptions, but unlikely.
• PIs are not obligated to engage the DSI in their future external grant applications. While involvement is possible and subject to mutual agreement, there is no expectation for it.

PI’s Trainee:

• Preference would be given to graduate student trainees over undergraduate student trainees, primarily because graduate students typically have greater availability to dedicate to the project for an extended duration compared to undergraduate students.
• DSI will not offer additional financial support for the trainees in the PI’s lab.

Thumbnail

Disentangling butterfly-plant associations across Canada using Machine Learning and community-contributed data

Michelle Tseng, Department of Botany

Butterfly populations are declining globally due to pesticides and habitat loss. Fortunately, the public's interest in butterflies has led to millions of photos on platforms like iNaturalist, which are now being used to track seasonal activity and distribution patterns. This project aims to develop algorithms to analyze these photos, identifying nectar plants and assessing their impact on butterfly ecology in British Columbia.

Poster presented during Poster Session in April 2026

Bromeliad Working Group integrated repository, packages and apps

Diane Srivastava, Department of Zoology

The distribution of energy through food webs, from prey to predator levels, forms biomass pyramids that depict biological energy stocks. Understanding the factors shaping these pyramids is hindered by limited comprehensive datasets across diverse climates and regions. The Bromeliad Working Group, led by Diane Srivastava at the University of British Columbia, has gathered ecological data from aquatic bromeliad food webs across the Americas over 15 years. They aim to use advanced data science to integrate and expand their database, enhancing their ability to analyze biomass pyramid shapes and their ecological drivers globally.

Poster presented during Poster Session in April 2026

Leveraging computer vision technology to monitor dynamic large mammal communities in Mozambique

Kaitlyn Gaynor, Department of Zoology

This project explores how human activity affects wildlife, particularly in Mozambique's Gorongosa National Park, where large mammals are recovering after civil conflict. Using motion-activated cameras since 2016, the research team tracks animal populations and their response to climate change and human pressures. Through the Data Science Catalyst Program, the team aims to create a database and computer vision tools to process images more efficiently and explore wildlife dynamics and the impact of disturbances like flooding and fires.

Poster presented during Poster Session in April 2026

Transfer learning for understanding seed dormancy

Liang Song, Department of Botany

Seeds are vital to both our society and ecosystems, with wild plants and crops differing in how their seeds behave. Wild plants use a strategy to cope with unpredictable environments by having seeds with varying dormancy levels, while crops have been bred for minimal dormancy. Understanding how seeds transition to seedlings is important, especially with climate change and extreme weather patterns. This research focuses on identifying the genetic factors that control seed dormancy and developing tools to predict how these factors work across different plant species.

Poster presented during Poster Session in April 2026

A Comparison-Driven Lameness Assessment Platform for Dairy Cows

Daniel Weary and Marina von Keyserlingk, Faculty of Land and Food Systems - with Tiffany Timbers in the Department of Statistics and Tamara Munzner in the Department of Computer Science

Lameness causes pain to dairy cows and economic losses to famers, but can be difficult to detect and routinely monitor. A new method using pairwise video comparisons and Bayesian algorithms offers a more reliable, scalable to way to assess lameness, outperforming traditional expert scoring. It also shows promise for broader use in fields like AI evaluation and medical diagnostics, though challenges remain in automation and interface design.

Poster presented during Poster Session in April 2026

From Farm to Insight: Scaling Data-Driven GHG Emissions Modelling and Nutrient Management Analytics Across the LiteFarm Global Platform

Hannah Wittman, Institute for Resources, Environment and Sustainability & Centre for Sustainable Food Systems in the Faculty of Land and Food Systems

LiteFarm is an open-source, participatory farm management platform developed at UBC serving over 9,000 farms across 160+countries, with deep datasets on crop management, harvest yields, nutrient applications, and livestock practices. Despite the richness of this data, a critical gap remains: translating raw farm management records into actionable, model-driven environmental insights; particularly for greenhouse gas (GHG) emissions and nutrient management at scale. This project proposes to close that gap by building a robust, integrated data science pipeline that connects LiteFarm’s global farm database to established GHG models, developing scenario-based decision support tools, and creating dynamic dashboards that update in real-time as model parameters evolve.

Defining Metabolic and Transcriptional Networks Linking Muribaculaceae to Clostridia-driven Gut Defense

Carolina Tropini, Department of Microbiology and Immunology

Loss of the commensal bacterial family Muribaculaceae (MB) profoundly alters gut ecosystem function and compromises resistance to pathogens, in part through disruption of microbiota members that are essential for gut function, the Clostridia. Our preliminary data demonstrate that MB supports the engraftment of specific Clostridia and restoration of key pathogen resistance metabolites such as butyrate. However, the molecular and ecological mechanisms underlying these interactions remain unresolved due to the complexity of multi-omic datasets generated from in vivo and in vitro models. In this project, we will leverage an integrated data science framework to combine metabolomics, metagenomics, and host and microbial transcriptomics from defined and complex microbial communities to identify pathways and networks mediating MB–Clostridia interactions.