<https://www.biorxiv.org/content/10.1101/2025.03.2...
# meta-omics
j
Pretty much validates hostile, right @Bede Constantinides?
b
Thanks for mentioning @James Fellows Yates, yes it’s an interesting and reassuring read. I’m less convinced by the blanket recommendation to use
--very-sensitive-local
however. I’ve no doubt that it performed best for their benchmark set, but against a much larger panel of simulated virus genomes I saw a major increase in false positive rate using more sensitive bowtie2 presets like this one. It all comes down to which taxa you’re dealing and how much you care about precision – sensitive presets will certainly remove more human reads, but this comes at a cost. For the record, you can pass sensitivity presets to Hostile should you wish with
--aligner-args
👍 3