Bayesian logistic regression with optional EX/NEX modeling enables flexible borrowing from historical or concurrent data. The safety model supports dose-escalation in adaptive Phase I oncology trials with multiple drugs. See Neuenschwander et al. (2008, 2016) for methodology.

This project demonstrates how to train a model that can generate synthetic brain (T1w) MRI data. The work was presented as a poster at ICAART 2025. Our code trains a Hierarchical Amortized GAN (HAGAN) model for 3D high resolution medical image synthesis of healthy subjects in MNI space.

tinydenseR is a landmark-based platform for single cell data analysis that identifies differentially abundant cell types and differentially expressed features, including subtle within-cluster state changes. Modeling samples as replicates, tinydenseR enhances analytic efficiency and reproducibility while preserving the richness of single cell data

singeCellComic is an interactive Shiny application designed to teach the fundamental concepts of single-cell RNA sequencing (scRNA-seq) analysis through a fun, story-driven, and visual narrative.

A Python library for benchmarking uncertainty estimation and quantification methods for Machine Learning models predictions.

Deep survival analysis made easy

Cellxgene Gateway allows you to use the Cellxgene Server provided by the Chan Zuckerberg Institute (https://github.com/chanzuckerberg/cellxgene) with multiple datasets.

Bayesian subgroup shrinkage estimation benchmark comparing various shrinkage models.

This pipeline is designed for AAV integration site analysis using Target Enrichment Sequencing (TES) coupled with PacBio long-read sequencing. It enables the detection of integration site breakpoints and provides information on the structure of the integrated AAV vector genome to support the safety assessment of AAV-based gene therapies.

Parse a simplified JSONPath-like syntax