Abstract:The increasing availability of large-scale single-cell datasets has enabled the detailed description of cell states across multiple biological conditions and perturbations. In parallel, recent advances in unsupervised machine learning, particularly in transfer learning, have enabled fast and scalable mapping of these new single-cell datasets onto reference atlases. The resulting large-scale machine learning models however often have millions of parameters, rendering interpretation of the newly mapped datasets challenging. Here, we propose expiMap, a deep learning model that enables interpretable reference mapping using biologically understandable entities, such as curated sets of genes and gene programs. The key concept is the substitution of the uninterpretable nodes in an autoencoder’s bottleneck by labeled nodes mapping to interpretable lists of genes, such as gene ontologies, biological pathways, or curated gene sets, for which activities are learned as constraints during reconstruction. This is enabled by the incorporation of predefined gene programs into the reference model, and at the same time allowing the model to learn de novo new programs and refine existing programs during reference mapping. We show that the model retains similar integration performance as existing methods while providing a biologically interpretable framework for understanding cellular behavior. We demonstrate the capabilities of expiMap by applying it to 15 datasets encompassing five different tissues and species. The interpretable nature of the mapping revealed unreported associations between interferon signaling via the RIG-I/MDA5 and GPCRs pathways, with differential behavior in CD8+ T cells and CD14+ monocytes in severe COVID-19, as well as the role of annexins in the cellular communications between lymphoid and myeloid compartments for explaining patient response to the applied drugs. Finally, expiMap enabled the direct comparison of a diverse set of pancreatic beta cells from multiple studies where we observed a strong, previously unreported correlation between the unfolded protein response and asparagine N-linked glycosylation. Altogether, expiMap enables the interpretable mapping of single cell transcriptome data sets across cohorts, disease states and other perturbations. ### Competing Interest Statement Fabian J. Theis consults for Immunai Inc., Singularity Bio B.V., CytoReason Ltd, and Omniscope Ltd, and has ownership interest in Dermagnostix GmbH and Cellarity.

Biophysical models of cis-regulation as interpretable neural networks

A mechanistically interpretable neural network for regulatory genomics

Interpreting cis -regulatory mechanisms from genomic deep neural networks using surrogate models

Interpreting cis-regulatory interactions from large-scale deep neural networks

Principled feature attribution for unsupervised gene expression analysis

Deciphering regulatory architectures from synthetic single-cell expression patterns

Deciphering gene regulation from gene expression dynamics using deep neural network

Biologically-informed neural networks guide mechanistic modeling from sparse experimental data

Learning the Regulatory Code of Gene Expression

Biologically Informed Deep Learning to Infer Gene Program Activity in Single Cells

Decoding biology with massively parallel reporter assays and machine learning

Interpreting -regulatory mechanisms from genomic deep neural networks using surrogate models

Modeling the Causal Regulatory Network by Integrating Chromatin Accessibility and Transcriptome Data.

Interpretability of Neural Network With Physiological Mechanisms

Biological Factor Regulatory Neural Network

Perspective on recent developments and challenges in regulatory and systems genomics

Hypothesis-driven interpretable neural network for interactions between genes

Synthesising Executable Gene Regulatory Networks from Single-cell Gene Expression Data

Dissecting endogeneous genetic circuits from first principles

Advances in computational and experimental approaches for deciphering transcriptional regulatory networks: Understanding the roles of cis-regulatory elements is essential, and recent research utilizing MPRAs, STARR-seq, CRISPR-Cas9, and machine learning has yielded valuable insights

Advancing regulatory genomics with machine learning