Finding insert sequences using long reads

Question

[bvm]

I'd like to find insert sequences on a known plasmid using Nanopore long reads. My first idea was to create a reference mapping to the plasmid with minimap2, then use the unmapped reads for polishing, but it didn't went well. Do you have any suggestion for a working pipeline?

[ATpoint]

You have a plasmid and something was inserted, is this the setup here? Details please.

[bvm]

I have a ~3k bp long plasmid and an insertion of the same length.

[ATpoint]

Wouldn't it be much simpler to design like 10 primers, every like 300bp of the known sequence and just Sanger that? At some point you will hit the insert for sure.

What rpolicastro says

[rpolicastro]

There are vendors which do nanopore plasmid seq for $15 a sample, so it's actually easier these days to nanopore plasmids.

[Anand Rao]

Thanks for this recommendation.

Based on your experience, what is consensus sequence accuracy for these ONT-based whole plasmid sequencing runs, and how is a low error rate achieved via fold coverage?

The technical team at a sequencing service company did not seem to think consensus seq. accuracy > raw seq. accuracy, despite their 200X coverage! And different team members quote 99.3% vs 99.9% accuracy, even after I shared these 2 links with them: ONT accuracy website and Mycota blog.

Error rate of ~ <= 1-2bp in 1 Sanger Sequencing run of ~ 800bp is what I am used to, and I would prefer not to deal with a higher error rate. I have never used ONT, hence this request for clarifications about ONT accuracy for just this specific application - whole plasmid sequencing.

Thanks in advance rpolicastro

[rpolicastro]

You can probably just do a de novo assembly with something as simple as minimap2 + miniasm.

[GenoMax]

See @Brian's answer here: Identification of the sequence insertion site in the genome

Answer 1

[colindaven]

Other suggestions are good, read mapping is not such a good idea.

I'd use flye or Raven to create a de novo assembly and then blast your inserts against that. Should be quite fast and easy. It's worked for me using this technique to find inserts in whole genomes.