The sourmash software package uses MinHash-based sketching to create “signatures”, compressed representations of DNA, RNA, and protein sequences, that can be stored, searched, explored, and taxonomically annotated. sourmash signatures can be used to estimate sequence similarity between very large data sets quickly and in low memory, and can be used to search large databases of genomes for matches to query genomes and metagenomes. sourmash is implemented in C++, Rust, and Python, and is freely available under the BSD license at http://github.com/dib-lab/sourmash.

No abstract

Single-cell RNA-seq (scRNA-seq) is a powerful tool for cell type identification but is not readily applicable to organisms without well-annotated reference genomes. Of the approximately 10 million animal species predicted to exist on Earth, >99.9% do not have any submitted genome assembly. To enable scRNA-seq for the vast majority of animals on the planet, here we introduce the concept of “k-mer homology,” combining biochemical synonyms in degenerate protein alphabets with uniform data subsampling via MinHash into a pipeline called Kmermaid. Implementing this pipeline enables direct detection of similar cell types across species from transcriptomic data without the need for a reference genome. Underpinning Kmermaid is the tool Orpheum, a memory-efficient method for extracting high-confidence protein-coding sequences from RNA-seq data. After validating Kmermaid using datasets from human and mouse lung, we applied Kmermaid to the Chinese horseshoe bat (Rhinolophus sinicus), where we propagated cellular compartment labels at high fidelity. Our pipeline provides a high-throughput tool that enables analyses of transcriptomic data across divergent species’ transcriptomes in a genome- and gene annotation-agnostic manner. Thus, the combination of Kmermaid and Orpheum identifies cell type-specific sequences that may be missing from genome annotations and empowers molecular cellular phenotyping for novel model organisms and species.