Genome assembly - 2 similar samples - one good, one bad

[Elsie]

Hi, currently doing genome assemblies on 2 very similar samples.

- 1 assembled brilliantly very quickly.
- 1 is highly fragmented

Any reason why one sample should behave so differently from another? - same sequence (Illumina HiSeq), same heterozygosity and repeat content, both screened for contaminants, both collected together, FastQC very similar for both, same assembly methodology, adapter trimmed.

Thoughts:
- adapters in the middle of reads?
- could a virus have inserted itself?

Any comments welcomed.

[Smurali]

Hmm.. interesting.

Some thoughts:
1. Was one more inbred than the other perhaps?
2. Did you check the insert sizes of the libraries? I'm thinking perhaps the mate pair library for the poor assembly resulting one wasn't as good as the other one.
3. Also, I have seen adapters in the middle of reads. You can quickly check for this if you know the adapter sequence.
4. I'm thinking if there was a virus, the virus sequence's kmer coverage would've been high enough for the assembler (de-bruijn graph based ones) to screen it out.

[Elsie]

Thanks Smurali.

These are field samples, not inbred. Everything is pointing towards a virus being integrated into the chromosome. Will do some more work on this and if I find out anything useful, will add another post.

[Brian Bushnell]

Thoughts:
- adapters in the middle of reads?
- could a virus have inserted itself?

Any comments welcomed.

Even if there were adapters in the middle of reads, they still would have been trimmed. And I don't see why a virus would cause a poor assembly, unless it randomly inserted itself into a different place in every cell. If it inserted itself once, then the cell replicated, you'd still get a good assembly.

It sounds more like cancer to me (depending on the organism), or degraded DNA. Have you looked at the insert size distribution and actual error rates of mapped reads (as opposed to just the quality scores)? Also, what is the read length, target insert size, and specific Illumina platform (e.g. HS2500) and run mode, and what kind of organism is it? Diploid or haploid? ...etc.

[Elsie]

Thanks for the comments Brian.
100bp PE, belong to the Hymenoptera order. Current evidence is pointing towards Polydnaviruses.

[Smurali]

Wow. this is certainly interesting.
I can only think of something external that somehow passed through your contamination screening and made it to the sequencing so the virus looks highly possible here.
When we sequenced and assembled a bunch of arthropods before, the final assembly sometimes did have a lot of contamination from Homo sapiens (on blood feeders), plants and viruses, so this is expected.
However, I am still intrigued by the fact that it is causing the assembly to be so highly fragmented. Are you going to try and assemble after removing the reads belonging to the virus, Elsie?