Whether TTT supports assembly outputs from hifiasm

[zhangyixing3]

Hi,

Thank you for developing such an excellent tool!

I would like to ask whether TTT supports assembly graphs generated by hifiasm.
Is it possible to use hifiasm outputs (e.g., GFA) directly as input for TTT?

Thank you!

[Dmitry-Antipov]

Hi,
In general it should work, but there are multiple possible issues.

a) hifiasm's gfa is not based on the de bruijn graph.
We transform gfa to a graph with sequence on edges (to be able to look for eulerian paths instead of hamiltonian paths). This transformation is trivial for (multiplex) de bruijn graphs, but for general case of gfa we have to use some heuristics and add some links that are not present in the original graph. Those fake links mostly likely will not be used in the optimal path output b TTT, but it is not guaranteed.
b) TTT needs long (ONT) reads alignment to the graph.
ONT alignments to hifi-only graph are done inside verkko with graphaligner as an intermediate assembly step. With hifiasm's graph graphaligner still should work, but it was not tested and issues are possible.
c) We've noticed that coverage reported by hifiasm for the final graph is not reliable for the repetitive nodes. verkko has the same issue, but it can be bypassed by using hifi-only graph (which also allows us to reuse verkko's internal read-to-graph alignments). For hifiasm we didn't explore how bad is the coverage situation, and what can be done to fix it.

[zhangyixing3]

Thank you so so much for your detailed response! I got it.

[lcanesin]

Hi @zhangyixing3
Did you try it in any way? Could you share how was it?

[zhangyixing3]

@lcanesin Sorry for the late reply. I’ve been really busy and haven’t had a chance to try it yet.