Constructing a vg graph from MSA with a large number of sequences

Question

[jdru]

Hi all,

I have a multiple sequence alignment of ~ 26000 sequences, each around 16kb. I would like to construct a vg graph of this alignment with all sequences embedded as haplotypes. However I am finding that

vg construct -M *msa_path*

Is too slow with this input size. I am able to construct the graph without the embedded paths however. I can also split the input into a smaller number of sequences and construct graphs for each subset of the MSA, but in this case I have not been able to merge these smaller graphs by overlapping nodes. So I was wondering if there was a simple way to achieve this, or is it simply infeasible to construct a graph in this way?

Thank you!

Answer 1

[Jouni Sirén]

You could try using PGGB instead of vg construct. I'm not sure how well it will scale to a large number of short sequences, as it's primarily inteded for a smaller number of long sequences.

Answer 2

[jdru]

Thank you so much, I have a graph! For some reason there are three connected components when there should only be one but at least it's a start.