UMIs are a popular way to identify duplicate DNA/RNA reads caused by PCR amplification. This requires software for collapsing duplicate reads with the same UMI, while accounting for sequencing/PCR errors. This tool implements many efficient algorithms for orders-of-magnitude faster UMI deduplication than previous tools (UMI-tools, etc.), while maintaining similar functionality. This is achieved by using faster data structures with n-grams and BK-trees, along other techniques that are carefully implemented to scale well to larger datasets and longer UMIs. Users of UMICollapse have reported speedups from taking hours or days to run with a previous tool to taking only a few minutes with this tool with real datasets! doi 10.7717/peerj.8275.