#40 · Tools & infrastructure

Space-time trade-offs in k-mer filters, superstrings, and repertoire search

The common thread is explicit space-time trade-off engineering in k-mer data structures. Super Bloom groups filter entries by minimizer to enable faster, cache-efficient streaming queries with fewer false positives via findere. Pareto optimization of masked superstrings jointly trades superstring length against mask runs, reporting 12-19% better compression of pan-genome k-mer sets. TCRseek applies windowed k-mer embeddings with FAISS indexes and alignment reranking to approximate nearest-neighbor search over TCR CDR3 repertoires, reporting 3.6-39.6x speedups over its comparison. What is new in this set is the explicit Pareto formulation of superstring compression and the minimizer-grouped filter layout for cache-friendly lookup. All three items are preprints.