Statistical Modeling
Software
AutoParts
A nonparametric Bayesian model that automatically infers the partition schemes for phylogenetic analyses of DNA data.
Manuscript is forthcoming
Open repository
Analysis
Cherry Blossom Bloom Forecaster
A competition-winning forecasting pipeline for predicting peak cherry blossom bloom with scarce data.
Read the write-upAnalysis
The Bayesian Battleship Benchmark
A Benchmark of LLM sequential decision-making against Bayesian policies.
Read the write-upOpen repository
Publication/Software
OmegaSwitch
An inference engine for non-parametric Bayesian models that detect changes in latent evolutionary regimes in codon datasets. Includes automatic model selection via RJ-MCMC.
Read the preprintOpen repository
Publication/Software
Inferring Gene Presence in Incomplete Data via Phylogenetic Occupancy Modeling
A probabilistic graphical model for imputing missing genes in large, highly-incomplete, genomic datasets.
Read the preprintOpen repository
Software
Smc-CAT
A C++ sequential Monte Carlo sampler for the CAT model, an important nonparametric Bayesian phylogenetic model.
Open repositoryBioinformatics
Publication
Comparative genomomic analysis of a chloroplast-stealing microorganism
Analyzed the genome of a kleptoplastic microorganism, with a focus on identifying and evaluating putative horizontal gene transfers that may support the acquisition and maintenance of stolen chloroplasts.
Read the preprintPublication
The nuclear and mitochondrial genomes of Amoebophrya
Assembly and comparative analysis of a marine dinoflagellate parasite, including characterization of its alternative genetic code and unusual mitochondrial genome architecture.
Read the paperMy Current Work
Machine learning
Graph neural networks for gene presence prediction
Graph neural networks for link prediction between genes and genomes, drawing inspiration from social recommendation systems.
Statistical modeling
Gene presence imputation with shared latent histories
A probabilistic clustering model that jointly infers shared gene histories to impute missing genes in partially observed genomes.