Andy Jones will present his Pre FPO "Probabilistic models for structured biomedical data" on May 26, 2022 at 10am via Zoom. Zoom link: https://princeton.zoom.us/j/91964333824 Title Probabilistic models for structured biomedical data Abstract It is commonly assumed in biomedical data analysis that the samples are independent and identically distributed. However, this assumption is nearly always violated in practice. Data are often composed of subgroups of samples (e.g., patients, cells, or tissue types) and usually contain substantial correlation structure among the samples and features. In this body of work, we propose a series of data modeling and analysis frameworks to account for these highly-structured experimental data. First, we propose two methods for performing dimension reduction and exploratory data analysis in a case-control setting. Second, we propose a spatial alignment method that registers spatial genomics slices into a common coordinate frame. Finally, in ongoing work, we propose a family of methods for optimally designing spatial genomics experiments. In each of these settings, we demonstrate the benefits of exploiting known experimental structure for data analysis and downstream findings. References Published or under review: Alignment of spatial genomics and histology data using deep Gaussian processes. Andrew Jones, F. William Townes, Didong Li, Barbara E. Engelhardt (2022). Multi-group Gaussian Processes. Didong Li, Andrew Jones, Sudipto Banerjee, Barbara E. Engelhardt (2021). Contrastive latent variable modeling with application to case-control sequencing experiments. Andrew Jones, F. William Townes, Didong Li, Barbara E. Engelhardt. The Annals of Applied Statistics (to appear). Nested policy reinforcement learning. Aishwarya Mandyam, Andrew Jones, Krzysztof Laudanski, Barbara E. Engelhardt (2021). Efficient Bayesian Inverse Reinforcement Learning via Conditional Kernel Density Estimation. Aishwarya Mandyam, Didong Li, Diana Cai, Andrew Jones, Barbara E. Engelhardt Probabilistic Contrastive Principal Component Analysis. Didong Li*, Andrew Jones*, Barbara E. Engelhardt (2020). In preparation: Optimizing experimental design for spatial transcriptomics studies. Andrew Jones, Barbara E. Engelhardt. Spatially-aware dimension reduction via semi-supervised latent variable models. Lauren Okamoto, Andrew Jones, Archit Verma, Barbara E. Engelhardt. A Poisson reduced rank regression model for association mapping in sequencing data. Tiana Fitzgerald, Andrew Jones, Barbara E. Engelhardt. Linking histology and molecular state across human tissues. Andrew Jones, Gregory W. Gundersen, Barbara E. Engelhardt.