Low coverage MinION reads should be useful to close gaps and resolve repeats left by short-read assemblers. However, I haven't had any success with the software I know about. I'm aware of the following packages, either for scaffolding or closing gaps in short-read assemblies using long reads:

I've tried npScarf and LINKS, and they ran successfully but didn't resolve the gaps in my assembly. I couldn't get PBJelly and OPERA-LG to run with the current versions of BLASR and samtools, and both packages seem not to be maintained. I have not tried SSPACE-longread because it's not open source.

What software can I use to fill gaps and resolve repeats in a short-read assembly with low-coverage Nanopore data?

More information

I'm finishing a mitochondrial genome. I have a 17 kb short read assembly with one gap. This was made from 2x150b paired-end reads from a TruSeq PCR-free library with a 400 b insert size. Molecular data suggests the genome may be around 32 kb.

After mapping short reads against the short read assembly, Pilon identified a tandem repeat of 156 bp in an AT-rich region, but wasn't able to close the gap.

I have 1.4 Gb of MinION rapid reads with an L50 of 8808 b. I mapped these reads against the short read assembly, and I can see reads that span the gap. I seem to have the information required to close the gap, but I don't know how to do it.

The organism has a nuclear genome > 600 Mb, so long read coverage is low. Despite this, I tried a de novo Canu assembly of the Nanopore reads, and I pulled out a 39 kb contig containing mitochondrial genes. Most genes on this contig are duplicated and fragmented, and Pilon wasn't able to improve it.

Thanks for reading!

  • $\begingroup$ Please provide some more details (single/pe reads, read length, insert size distribution if pe, sequencing technology) about your short read assembly. If you have overlapping paired-end Illumina reads, I suggest to merge them and try to improve the assembly afterwards. Your short read assembly is about half of the size what you expect from molecular data (which kind of data). So maybe, there is a larger duplication. I'm not familiar with mito genomes and don't know if duplications exist in general in mito genomes. Neither I know the organism you try to assemble. Did you see any region(s) with $\endgroup$ – Mathias Walter Jul 26 '17 at 10:24
  • $\begingroup$ Thanks for the comment. I added SR lib details. There are several possibilities for the the size discrepancy. It could relate to a duplication, nuclear insertion of mtDNA, a repeat region etc. It's also possible the molecular data (RE mapping and Southern blots) could be wrong. It would be helpful to close the gap! $\endgroup$ – Tom Harrop Jul 26 '17 at 10:44

You might want to look into Unicycler (manuscript with more information can be found here); even though it is supposed to be used with bacterial genomes only, it might work well with a small genome such as a mitochondrion's.

enter image description here

Beware that if you happen to have very long reads, you might end up with an assembly with multiple copies of the circular genome: you might want to look into circlator then.

| improve this answer | |
  • $\begingroup$ Mitochondria are bacteria. Their long co-evolution as endosymbiotes may have had an influence in some of their genomic characteristics, though. $\endgroup$ – bli Jul 26 '17 at 11:36
  • $\begingroup$ Exactly; in fact, I'm not aware of any bacteria with a ~32kb genome :) $\endgroup$ – mgalardini Jul 26 '17 at 12:44
  • 1
    $\begingroup$ This worked well for me. I got a single 21 kb circular contig with all the expected mt genes. Thanks for alerting me to this software! $\endgroup$ – Tom Harrop Jul 31 '17 at 23:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.