I am analysing 125bp paired-end metagenomic data, using kraken2 for taxonomic classification. I am running into memory issues, since I am using a very large custom Database. The size of the database is significantly decreased without removing sequences, when I increase the k-mer length. Currently i am using a value of 43 instead of the default of 35. I don't know how this is going to affect my results. The documentation does not give any clear information about this issue. Is it save to increase this value or should I rather remove some sequences to reduce the database size?