How to assemble long reads to scaffolds?

[dacotahm]

I would like to use some long reads from new nanopore data to lengthen and bridge scaffolds in an existing genome release. Does anyone have a recommendation for an assembler that would work well or is built for this task and tolerate (or preferably fill in) the NNNNs in the existing scaffolds?

[fahmida]

Typically due to high error rate you have to correct your nanopore reads by more accurate short reads from e.g. illumina. There are several methods to do that. Then you can use something like SSPACE longread for scaffolding.

[gringer]

Typically due to high error rate you have to correct your nanopore reads by more accurate short reads

This is necessary for finishing a genome, but not for the initial scaffolding.You can use something like Miniasm to do a high-error scaffold, which may work if you just throw your current scaffolds in with nanopore reads.

Alternatively, depending on how big your genome is, you could do another de-novo assembly from scratch using Canu. People have been quite successful with Nanopore assemblies of bacterial and yeast genomes using Canu.

[dacotahm]

My plan was to use the long reads with high error rate to bridge scaffolds and then map Illumina on to them, swap bases to favor Illumina, and hopefully fill in some of the Ns

[n0rr]

canu http://biorxiv.org/content/early/2016/08/24/071282
canu can assemble mammalian genome with cheap ws. https://arxiv.org/abs/1703.10231