Yongin Choi’s research while affiliated with University of California, Davis and other places

A bstract Single-cell RNA sequencing (scRNA-seq) enables high-resolution exploration of cellular diversity and gene regulation, yet analyzing such data remains challenging due to technical and methodological limitations. Existing task-specific deep generative models like Variational Auto-Encoder (VAE) and its variants struggle to incorporate external biological knowledge, while transformer-based foundational large Language Models (LLMs or large LaMs) face limitations in computational cost and applicability to tabular gene expression data. Here, we introduce sciLaMA (single-cell interpretable Language Model Adapter), a novel representation learning framework that bridges these gaps by integrating static gene embeddings from multimodal LaMs with scRNA-seq tabular data through a paired-VAE architecture. Our approach generates context-aware representations for both cells and genes and outperforms state-of-the-art methods in key single-cell downstream tasks, including batch effect correction, cell clustering, and cell-state-specific gene marker and module identification, while maintaining computational efficiency. sciLaMA offers a computationally efficient, unified framework for comprehensive single-cell data analysis and biologically interpretable gene module discovery.

Here we present biVI, which combines the variational autoencoder framework of scVI with biophysical models describing the transcription and splicing kinetics of RNA molecules. We demonstrate on simulated and experimental single-cell RNA sequencing data that biVI retains the variational autoencoder’s ability to capture cell type structure in a low-dimensional space while further enabling genome-wide exploration of the biophysical mechanisms, such as system burst sizes and degradation rates, that underlie observations.

The Ras/ERK pathway drives cell proliferation and other oncogenic behaviors, and quantifying its activity in situ is of high interest in cancer diagnosis and therapy. Pathway activation is often assayed by measuring phosphorylated ERK. However, this form of measurement overlooks dynamic aspects of signaling that can only be observed over time. In this study, we combine a live, single-cell ERK biosensor approach with multiplexed immunofluorescence staining of downstream target proteins to ask how well immunostaining captures the dynamic history of ERK activity. Combining linear regression, machine learning, and differential equation models, we develop an interpretive framework for immunostains, in which Fra-1 and pRb levels imply long term activation of ERK signaling, while Egr-1 and c-Myc indicate recent activation. We show that this framework can distinguish different classes of ERK dynamics within a heterogeneous population, providing a tool for annotating ERK dynamics within fixed tissues.

Neural networks such as variational autoencoders (VAE) perform dimensionality reduction for the visualization and analysis of genomic data, but are limited in their interpretability: it is unknown which data features are represented by each embedding dimension. We present siVAE, a VAE that is interpretable by design, thereby enhancing downstream analysis tasks. Through interpretation, siVAE also identifies gene modules and hubs without explicit gene network inference. We use siVAE to identify gene modules whose connectivity is associated with diverse phenotypes such as iPSC neuronal differentiation efficiency and dementia, showcasing the wide applicability of interpretable generative models for genomic data analysis.

We motivate and present biVI, which combines the variational autoencoder framework of scVI with biophysically motivated, bivariate models for nascent and mature RNA distributions. In simulated benchmarking, biVI accurately recapitulates key properties of interest, including cell type structure, parameter values, and copy number distributions. In biological datasets, biVI provides a route for the identification of the biophysical mechanisms underlying differential expression. The analytical approach outlines a generalizable strategy for representing multimodal datasets generated by single-cell RNA sequencing.

Deep neural networks implementing generative models for dimensionality reduction have been extensively used for the visualization and analysis of genomic data. One of their key limitations is lack of interpretability: it is challenging to quantitatively identify which input features are used to construct the embedding dimensions, thus preventing insight into why cells are organized in a particular data visualization, for example. Here we present a scalable, interpretable variational autoencoder (siVAE) that is interpretable by design: it learns feature embeddings that guide the interpretation of the cell embeddings in a manner analogous to factor loadings of factor analysis. siVAE is as powerful and nearly as fast to train as the standard VAE but achieves full interpretability of the embedding dimensions. We exploit a number of connections between dimensionality reduction and gene network inference to identify gene neighborhoods and gene hubs, without the explicit need for gene network inference. Finally, we observe a systematic difference in the gene neighborhoods identified by dimensionality reduction methods and gene network inference algorithms in general, suggesting they provide complementary information about the underlying structure of the gene co-expression network.

Intratumoral heterogeneity is associated with aggressive tumor behavior, therapy resistance, and poor patient outcomes. Such heterogeneity is thought to be dynamic, shifting over periods of minutes to hours in response to signaling inputs from the tumor microenvironment. However, models of this process have been inferred from indirect or post-hoc measurements of cell state, leaving the temporal details of signaling-driven heterogeneity undefined. Here, we developed a live-cell model system in which microenvironment-driven signaling dynamics can be directly observed and linked to variation in gene expression. Our analysis reveals that paracrine signaling between two cell types is sufficient to drive continual diversification of gene expression programs. This diversification emerges from systems-level properties of the EGFR-RAS-ERK signaling cascade, including intracellular amplification of amphiregulin-mediated paracrine signals and differential kinetic filtering by target genes including Fra-1, c-Myc, and Egr1. Our data enable more precise modeling of paracrine-driven transcriptional variation as a generator of gene expression heterogeneity. A record of this paper’s transparent peer review process is included in the Supplemental Information.