Accelerating k -mer-based sequence filtering

The exponential growth of global sequencing data repositories presents both analytical challenges and opportunities. While k - mer-based indexing has improved scalability over traditional alignment for identifying potentially relevant documents, pinpointing the exact sequences matching a query remains a hurdle, especially when considering the false positives and false negatives caused by sparse k -mer sampling or probabilistic data structures. Furthermore, searching for numerous k -mers with a large query or multiple distinct ones strains existing exact matching tools, whose performance scales poorly with an increasing number of patterns. Indexing entire vast datasets for infrequent or adhoc searches is often resource-prohibitive. This paper addresses the renewed need for efficient streaming techniques to recover specific relevant sequences directly, without exhaustive pre-indexing. Having discussed the relevance of streaming search methods in bioinformatics, we propose diverse use cases as benchmarks to encourage further research into tailored solutions for these challenges, and demonstrate how sketching techniques like minimizers, alongside SIMD acceleration, can enhance the performance of such streaming searches. These contributions highlight practical approaches and identify avenues for future theoretical and empirical investigation in managing and querying vast genomic datasets. Availability: https://github.com/Malfoy/K2Rmini .