ACE: A Versatile Contrastive Learning Framework for Single-cell Mosaic Integration

TL;DR

ACE is a new framework that improves the integration of single-cell multi-omics data by aligning and completing missing modalities across datasets.

Contribution

ACE introduces two novel strategies, ACE-align and ACE-spec, for mosaic integration of single-cell data using contrastive learning and regression.

Findings

ACE-align uses contrastive learning to align modalities and reveal shared latent representations.

ACE-spec enhances cellular heterogeneity representation in datasets with incomplete modalities.

ACE outperforms existing methods in various mosaic integration scenarios.

Abstract

The integration of single-cell multi-omics datasets is critical for deciphering cellular heterogeneities. Mosaic integration, the most general integration task, poses a greater challenge regarding disparity in modality abundance across datasets. Here, we present Align and CompletE (ACE), a mosaic integration framework that assembles two types of strategies to handle this problem: modality alignment-based strategy (ACE-align) and regression-based strategy (ACE-spec). ACE-align utilizes a novel contrastive learning objective for explicit modality alignment to uncover the shared latent representations behind modalities. ACE-spec combines the modality alignment results and modality-specific representations to construct complete multi-omics representations for all datasets. Extensive experiments across various mosaic integration scenarios demonstrate the superiority of ACE’s two strategies over existing methods. Application of ACE-spec to bi-modal and tri-modal integration scenarios showcases that ACE-spec is able to enhance the representation of cellular heterogeneities for datasets with incomplete modalities. The source code of ACE can be accessed at https://github.com/CSUBioGroup/ACE-main.

Figures