Caleb N. Ellington

GHC 7405

Gates Hillman Complex, 4902 Forbes Ave

Pittsburgh, PA 15213

I work on AI/ML for biology to enable automated biomarker discovery, drug repurposing, and therapeutic target identification for heterogeneous and rare diseases. My research interests span foundation models, probabilistic graphical models, biological modeling, and multi-task learning.

I co-created contextualized.ml, an sklearn-style toolbox for personalized modeling with heterogeneous observational data. Contextualized Machine Learning is a novel machine learning paradigm that estimates models, functions, and distributions at sample-specific resolution, allowing per-patient or per-cell analysis of biomedical data and enabling truly personalized modeling for healthcare, precision diagnostics, and drug repurposing.

I also enjoy rock climbing, running, backpacking, guitar, learning languages, and wild speculation about Spotify’s recommender system. I’ve lived in Austin TX, Seattle WA, and Pittsburgh PA and I call them all home.

news

Dec 4, 2023	Contextualized Networks Reveal Heterogeneous Transcriptomic Regulation in Tumors at Sample-Specific Resolution is online now on bioRxiv. We introduce a new machine learning paradigm for sample-specific inference of probabilistic graphical models, and show that sample-specific models of transcriptional regulation are highly accurate, interpretable, and prognostic across 25 tumor types and nearly 8000 patints. I’ll be sharing significant updates to the method and results at NeurIPS 2023 GenBio Workshop.
Oct 17, 2023	The Contextualized Machine Learning White Paper is online at arXiv here. We explore applications, algorithms, and extensions for contextualized models: models that understand heterogeneity in real data, adapt to new environments, and are explainable by design. This work outlines the vision for the Contextualized ML project, which I co-created with Ben Lengerich and has transformed into a fast-growing reasearch community.
Oct 11, 2023	Contextualized Policy Recovery: Modeling and Interpreting Medical Decisions with Adaptive Imitation Learning is online on arXiv here. This work is co-first with my extremely talented MS student Jannik Deuschel.
Apr 26, 2023	Contextualized ML is one year old today. Over the past year we’ve gained 50 GitHub stars and over 6000+ downloads.
Nov 16, 2022	Check out my talk from CSHL Biological Data Science this past week: Contextualized Graphical Models Reveal Sample-Specific Transcriptional Networks for 7000 Tumors

selected publications

arXiv

Contextualized Machine Learning

Lengerich, Benjamin, Ellington, Caleb N., Rubbi, Andrea and 2 more authors

Abs HTML

We examine Contextualized Machine Learning (ML), a paradigm for learning heterogeneous and context-dependent effects. Contextualized ML estimates heterogeneous functions by applying deep learning to the meta-relationship between contextual information and context-specific parametric models. This is a form of varying-coefficient modeling that unifies existing frameworks including cluster analysis and cohort modeling by introducing two reusable concepts: a context encoder which translates sample context into model parameters, and sample-specific model which operates on sample predictors. We review the process of developing contextualized models, nonparametric inference from contextualized models, and identifiability conditions of contextualized models. Finally, we present the open-source PyTorch package ContextualizedML.
bioRxiv

Contextualized Networks Reveal Heterogeneous Transcriptomic Regulation in Tumors at Sample-Specific Resolution

Ellington, Caleb N., Lengerich, Benjamin J., Watkins, Thomas BK and 4 more authors

Abs HTML

Cancers are shaped by somatic mutations, microenvironment, and patient background, each altering gene expression and regulation in complex ways, resulting in heterogeneous cellular states and dynamics. Inferring gene regulatory network (GRN) models from expression data can help characterize this regulation-driven heterogeneity, but network inference requires many statistical samples, traditionally limiting GRNs to cluster-level analyses that ignore intra-cluster heterogeneity. We propose to move beyond cluster-based analyses by using contextualized learning, a multi-task learning paradigm which allows us to infer sample-specific models using phenotypic, molecular, and environmental information pertinent to the model, encoded as the model’s "context" to be conditioned on. We unify three network model classes (Correlation, Markov, Neighborhood) and estimate context-specific GRNs for 7997 tumors across 25 tumor types, with each network contextualized by copy number and driver mutation profiles, tumor microenvironment, and patient demographics. Contextualized GRNs provide a structured view of expression dynamics at sample-specific resolution, which reveal co-expression modules in correlation networks (CNs), as well as cliques and independent regulatory elements in Markov Networks (MNs) and Neighborhood Regression Networks (NNs). Our generative modeling approach allows us to predict GRNs for unseen tumor types based on a pan-cancer model of how somatic mutations affect gene regulation. Finally, contextualized networks enable GRN-based precision oncology, explaining known biomarkers in terms of network-mediated effects, and leading to novel subtypings for thyroid, brain, and gastrointestinal tumors that improve survival prognosis.