How to replicate sourmash hash function

[Adamtaranto]

Hi Titus and co.,

I want to keep track of kmer locations in a genome so I can do some downstream visualisation of select kmers from a MinHash object. I'm currently extracting each kmer from the genome, calculating its canonical hash value and storing its observed location in an SQLite db so I can quickly lookup all locations of a kmer later.

I see that I can use the MinHash.seq_to_hashes() method to get the hash value for a particular kmer, but this feels kinda clunky.

If I use mmh3.hash64() with seed 42 as per compute-dna-mh-another-way.py to calculate the hash directly will this always return the same value for a given kmer as any MinHash object? or does MinHash make other conditional mods to the hash?

Thanks!

[ctb]

hi @Adamtaranto, great question!

The script you found is exactly the code we use internally (just, you know, faster). I would definitely apply "trust-but-verify" to my statement, but we put this code in place back in ~2016, built a bunch of tests around it, and haven't changed it since then - so 🤞 it better be the same!

[ctb]

ref #3339