2
$\begingroup$

I have assembled a virus genome using Ray resulting in approximately 5000 contigs. Now I want to build a scaffold of those contigs to get the full genome (I am aware of the fact that there might be some gaps in the genome).

I found one tool called Medusa to do this. Unfortunately, this tool cannot take a file above 50 MB and my contig file is about 156 MB.

Are there any other tools that can perform reference based scaffolding?

It would be great if anyone knows about genome scaffolding packages in Bioconda.

Improve this question
$\endgroup$
2
  • $\begingroup$ Any genomics tool that doesn't work over 50MB data is meaningless. Please delete Medusa. $\endgroup$
    – ABCD
    Jul 19, 2018 at 12:02
  • $\begingroup$ @SmallChess if it's designed for viral genomes, why not. $\endgroup$
    – Kamil S Jaron
    Jul 23, 2018 at 18:52

3 Answers 3

Reset to default
2
$\begingroup$

The number of contigs and total assembly size you have suggest that there was probably more in the sequencing run than a single virus strain. Does the total assembly length correspond to expected genome size of the reference strain? Have you tried just to blast some your contigs? Could you have a contamination or something?

Maybe you can just sort out the contigs by the reference using something like artemis or quast compares your assembly to reference and showing nice stats (just to be sure that your assembly makes sense so far). Or you could use directly MUMmer (which the alignment tool that is running on the backend of of Medusa and Quast).

This is not directly answering your question "How to scaffold", however I think you should first figure out if your contigs make sense before you scaffold them. Also sorting your contigs by artemis using reference is practically the same thing as scaffolding since you won't know the sequence in between in either of the cases.

Improve this answer
$\endgroup$
5
  • $\begingroup$ You are right. There is a lot of host genome contamination. I blasted all the contigs against reference genome and only about 10 contigs had significant match with the reference. I was looking for some tool that could take these 10 contigs, re-order them and join them together or add NNNNs if the sequence is missing between the contigs. $\endgroup$
    – L R Joshi
    Jul 23, 2018 at 21:28
  • $\begingroup$ And do you know the host genome? If yes, you could map everything to both host and the reference and then kick out everything that maps to host and scaffold just non-host contigs. $\endgroup$
    – Kamil S Jaron
    Jul 23, 2018 at 21:42
  • $\begingroup$ I know the host genome because it is cell-cultured virus. And I am actually looking for some tools that could actually do what you have mentioned. $\endgroup$
    – L R Joshi
    Jul 23, 2018 at 23:44
  • $\begingroup$ I would personally just run MUMmer of the assembly vs catinated reference of the host and related strain. MUMmer allows secondary alignments as well so you will be able to find also regions that are similar between the both, then parse the output in python/R/whateveryoulike to create a list of scaffolds you want keep/remove and use it to filter out our reference (a script to do that could look like this github.com/KamilSJaron/generic_genomics/blob/master/…). I should probably add this to the answer. $\endgroup$
    – Kamil S Jaron
    Jul 24, 2018 at 6:25
  • $\begingroup$ Thanks Kamil. I will try that script. I also found that Viral NGS Pipeline from broad institute can actually do this. There is a python script called order_and_orient which serves this purpose. Here is the link viral-ngs.readthedocs.io/en/latest/assembly.html $\endgroup$
    – L R Joshi
    Jul 26, 2018 at 2:19
1
$\begingroup$

What you mean is assembly based scaffolding, as opposed to using reads with long distance information such as mate pair/long jumping distance to scaffold.

You could reduce your contig file to just the longest contigs, since they have the most information. Probably 156MB of data is overkill for your virus.

There are other programs out there.

https://github.com/ksahlin/BESST

https://github.com/institut-de-genomique/MaGuS

Your mileage will vary.

Improve this answer
$\endgroup$
2
  • $\begingroup$ I tried that. I took five largest contigs. The output is not good. I will try the tools that you provided. Thank you. $\endgroup$
    – L R Joshi
    Jul 20, 2018 at 17:55
  • $\begingroup$ Isn't BESST only for read-based scaffolding? $\endgroup$
    – Kamil S Jaron
    Aug 14, 2018 at 13:22
1
$\begingroup$

I found that Viral NGS Pipeline from broad institute can actually do this. There is a python script called order_and_orient which serves this purpose. Here is the link viral ngs assembly

Improve this answer
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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