Multi-view clustering by CPS-merge analysis with application to multimodal single-cell data

Multi-view data can be generated from diverse sources, by different technologies, and in multiple modalities. In various fields, integrating information from multi-view data has pushed the frontier of discovery. In this paper, we develop a new approach for multi-view clustering, which overcomes the limitations of existing methods such as the need of pooling data across views, restrictions on the clustering algorithms allowed within each view, and the disregard for complementary information between views. Our new method, called CPS-merge analysis, merges clusters formed by the Cartesian product of single-view cluster labels, guided by the principle of maximizing clustering stability as evaluated by CPS analysis. In addition, we introduce measures to quantify the contribution of each view to the formation of any cluster. CPS-merge analysis can be easily incorporated into an existing clustering pipeline because it only requires single-view cluster labels instead of the original data. We can thus readily apply advanced single-view clustering algorithms. Importantly, our approach accounts for both consensus and complementary effects between different views, whereas existing ensemble methods focus on finding a consensus for multiple clustering results, implying that results from different views are variations of one clustering structure. Through experiments on single-cell datasets, we demonstrate that our approach frequently outperforms other state-of-the-art methods.Author summary: Advances in single-cell profiling technologies have made it possible to measure various types of features from a single cell. In this new type of data, known as multimodal single-cell data, each cell has numerical measurements from multiple views. Analyzing multimodal data has opened up new horizons for single-cell genomics, where clustering is a fundamental analysis for validating existing hypotheses or discovering insights when little prior knowledge is available. Existing clustering methods either combine data from different modalities for simultaneous processing or use integration algorithms to aggregate clustering results from multiple views. In this paper, we propose a new approach called CPS-merge analysis, which considers both consensus and complementary effects among clustering results across views and provides a quantified contribution of each view. The approach operates on single-view cluster labels, enabling the use of advanced clustering algorithms in any individual view. Furthermore, since CPS-merge analysis does not require pooling the original data, it can be applied to distributed sources or data with sharing concerns. This new approach tackles the problem of multi-view clustering from a novel combinatorial perspective and has the potential to become a widely used and effective tool.