VG Giraffe does not map reads < 39 bases

Question

[jdru]

Hello,

My colleagues and I have noticed that vg giraffe will not map reads of length 38 or shorter. I assume this was an intentional design decision but I was hoping to verify that this is indeed the case? And if so would there be any simple way to override this behavior?

If this is not known behavior I would be glad to post an example where we observe this.

Thank you,
Joshua

Answer 1

[anovak]

This would be caused by the k-mer size and window count of the minimizer index, which default to 29-mers and then the minimal one of them in 11 consecutive windows. If you have fewer than 39 bp you don't actually have 11 consecutive 29-mer sliding windows, and so your read cannot have any minimizers and thus has nothing to look up in the index.

If you make your minimizer index with vg minimizer with custom -k and -w settings, you can make a minimizer index that supports a smaller minimum read length.