cyto: ultra high-throughput processing of 10x-flex single cell sequencing

Single-cell genomics is rapidly scaling toward billion-cell atlases, but computational analysis has become a critical bottleneck. Processing multiplexed datasets with existing tools requires substantial computational resources and runtime that become prohibitive at scale. Here we present cyto, an ultra highthroughput processor for 10x Genomics Flex single-cell sequencing optimized for production-scale analysis. cyto exploits the fixed sequence geometry of Flex libraries through direct k-mer lookup rather than alignment-based mapping, and introduces IBU (Indexed-Barcode-UMI), a compact binary format for efficient read processing. cyto further leverages BINSEQ, a binary sequencing format that enables highly parallel parsing and overcomes the single-threaded limitations of gzip compression. On a benchmark 320,000-cell multiplexed dataset, cyto completes processing in 13 minutes compared to CellRangers 3.7 hours, a 16.5-fold speedup, while requiring 2.4-fold less memory and performing 5.6-fold less disk I/O. The 31.7-fold reduction in CPU-hours represents true algorithmic efficiency rather than parallelization alone. Critically, cyto maintains 99.85% concordance with CellRanger outputs, with identical cell type clustering in dimensionality reduction analyses. These performance improvements enable costeffective processing on smaller cloud instances and make previously prohibitive experiments computationally feasible. cyto is production-ready, open-source software that provides the computational foundation for atlas-scale single-cell projects and genome-wide perturbation screens.