Methods are now available that enable generation of very large scATAC-seq datasets8. This presents opportunities to deeply characterize the chromatin state of tissues at single-cell resolution, but also raises the need for computational tools that similarly scale to large cell numbers. To explore how the runtime of key analysis steps in the Signac framework scales to large numbers of cells, we analyzed two separate scATAC-seq datasets of differing sizes: a human PBMC scATAC-seq dataset of 26,579 cells from 10X Genomics and an adult mouse brain dataset of 734,000 cells from the Brain Initiative Cell Census Network (BICCN)53. We downsampled the full dataset down to 1,000 cells (for the PBMC dataset) or 50,000 cells (for the BICCN dataset) and ran each step in a full analysis workflow, including creating the initial object required, quantifying counts in peaks, quantifying DNA accessibility at each gene, computing QC metrics and performing dimensionality reduction. We also compared the runtime for Signac to the recently published ArchR package35 for equivalent analysis steps. For steps that are able to be run in parallel, we tested with 1,
