Regarding the node length requirements for using gfa files in vg giraffe

Question

[Wenhai]

Hello everyone,

I want to ask a simple question. In vg giraffe, how long is it best to control the node length in the gfa file? Will a node that is too long cause a crash during index building or alignment?

Thank you very much.

Answer 1

[Jouni Sirén]

The minimizer index requires that the nodes are no longer than 1024 bp. When the GFA file is converted to GBZ, the conversion will automatically chop nodes to a specific length (default 1024 bp) when necessary. If that happens, the node identifiers in the GBZ graph will not be the same as in the GFA graph.

Some vg tools have an option to convert the node identifiers in the output to those used in the original GFA graph. Unfortunately this is not universal yet. If you want a GFA file that is identical to the GBZ graph Giraffe uses, you can convert the GBZ back to GFA without using the node id translation: vg convert --gfa-out --no-translation graph.gbz > graph.gfa.

[Wenhai]

I got it. Thank you.