Large Language Models | Jennifer Yu

Generalist large language models complement tailor-made predictors for tumor genomics interpretation

Fri, 22 May 2026 00:00:00 +0000

Large language models for tumor genomic interpretation

Fri, 03 Apr 2026 00:00:00 +0000

Also presented at ISMB 2026.

CodeWearBook: Retrieval-Grounded Natural-Language Querying of Wearable Health Datasets Using Large Language Models

Thu, 01 Jan 2026 00:00:00 +0000

Generalist Large Language Models for Tumor Genomics Interpretation

Wed, 01 Jan 2025 00:00:00 +0000

Tumor genomic profiling produces results that must be interpreted before they can guide care: what is the most likely cancer type, is a variant somatic or the result of clonal hematopoiesis (CHIP), and is it likely to be oncogenic? Specialized, tailor-made predictors exist for each of these questions, but they are trained on a single institution’s data and labels and can degrade when deployed elsewhere.

In this project, which I lead as part of my PhD work with Dr. Quaid Morris and Dr. Justin Jee at Memorial Sloan Kettering Cancer Center, we benchmark zero-shot generalist large language models against supervised baselines on cancer-type, somatic-vs-CHIP and variant-oncogenicity classification using more than 100,000 samples from 19 institutions. We find that generalist LLMs and tailor-made predictors are complementary: LLM–specialist ensembles raise cross-institution accuracy by up to 55%. The work was presented at the AACR Annual Meeting 2026 and ISMB 2026, and a preprint is available on bioRxiv.

CodeWearBook: Agentic Natural-Language Querying of Wearable Health Datasets

Mon, 01 Jan 2024 00:00:00 +0000

Longitudinal wearable studies such as those run by the Goldenberg Lab in collaboration with 4YouandMe produce large, heterogeneous datasets of physiological signals, surveys and clinical events. Asking even simple questions of these data usually requires a data engineer and custom code.

CodeWearBook is an agentic workflow, developed with Dr. Anna Goldenberg’s group at the University of Toronto, the Vector Institute and SickKids, in which large language models answer natural-language questions about wearable health datasets by generating and executing code. The workflow is designed to be adaptable to new data sources and contexts while ensuring data privacy and security by restricting the LLM’s access to raw participant data through controlled SQL and Python execution layers. A manuscript describing the system is under review.