DIANA Reference K-mer Database v1.0

Reference k-mer database for the DIANA (Deep learning-based Identification and ANnotation of Ancient DNA samples) pipeline. Contains 18,953,119 discriminative k-mers (k=31) extracted from a training matrix of 3,070 ancient metagenome samples. This file is used as the reference for counting k-mers in new samples during inference, enabling rapid classification of ancient DNA samples without reprocessing the full training matrix. File: reference_kmers.fasta (~750 MB)Format: FASTAK-mer size: 31Total k-mers: 18,953,119