for estimating DNA motif variability between cells32, Monocle for building pseudotime trajectories54, Cicero for finding co-accessible networks of peaks31 and Harmony for performing dataset integration55. While we have focused here on the analysis of DNA accessibility data alone, our suite of tools is also fully compatible with recently developed methods to construct a joint neighbor graph encompassing multiple data modalities42, enabling chromatin data to be combined with additional modalities to jointly define cell states. As additional experimental methods for measuring multiple aspects of cell state are developed, a major challenge is to analyze these diverse datasets together in a consistent framework to learn how different modalities influence one another. The Seurat framework, via the extensible Assay class, is an appealing solution for the analysis of multimodal single-cell data and we envision future computation methods will further build on the Seurat and Signac frameworks to jointly analyze multimodal single-cell datasets. Future work will aim to further improve the scalability of the Signac package and to expand the suite of computational methods implemented. A major challenge currently facing biology is understanding how the genome encodes the organism56. Developing a deep understanding of how genes are regulated by noncoding DNA elements would greatly
