Hi,

one way you might want to look at this is by mapping your sequencing reads onto your scaffolds and look at the coverage pattern for the regions-of-concern and for the rest of the genome. If you observe a largely uniform pattern this might be a good indication of a genuine assembly.

Alternatively, given a close reference genome, you might just map the reads onto that reference and check the coverage pattern there if the reference genome has similar regions. Again, if you observer non-uniformity (e.g. a 2-fold coverage compared to the rest of the genome) this might give you an idea.

Best,

Cedric

Hi Cedric,

Thank you for your prompt reply. I have found transposable elements in that part of scaffolds. So can this be the reason for duplication?

I will surely map reads on those scaffolds and try to check as per your suggestion.

Regards,

Rajesh

Hi Rajesh,

this is a tricky question as we are going into a very detailed analysis. Informally, I think that observing transposons in these regions is a further indicator for a possible duplication; after all that is what transposons are known for right? ;)

Maybe the recent microseminar from Lizzy Wilbanks (https://www.youtube.com/watch?x-yt-cl=84359240&feature=player_detailpage&x-yt-ts=1421782837&v=zKvDEh-uPqc#t=1640) might be an interesting look at.

Also, I think it is important to not overestimate what one can do from a computational analysis. Once good candidates are identified with a "sufficient" degree of confidence (this depends on various factors and is an individual thing) one should think about going for wet-lab validation IMHO. Although I do not know how this would be done in your particular case. My guess, using (q)PCR and Sanger/long read sequencing.

Best,

Cedric

Hi Cedric,

Thanks for your suggestion. I also think estimating the copy number of genes in duplicated scaffold region using qPCR approach may help.I will further discuss this idea with my team.

Regards,

Rajesh

Hi,

do you mean that you have several contigs and two of these are 99.7% similar with the genes identical AND these two contigs belong two different scaffolds?

Or you have two contigs in which 25kb regions are so similar?

The problem is that the assembly programs usually can not decide which reads to put into which of the corresponding identical regions, hence the multicopy regions appear only once in the assembly (with appropriately higher coverage than the rest).

So the problem you encountered is at least strange, and I would suspect some technical error rather than duplication event.

Nevertheless, if you have paired-end data that proves that the region in fact appears twice in the genome, then it may still be OK. Still, depending on the accuracy of the sequences (quality data) and the coverage, the ~75 mismatches may or may not be significant.

Without knowing details such as chemistry, assembly program, coverage, etc., it is hard to judge.

Hi, when we sequenced the genome of Xanthomonas oryzae, a bacterial plant pathogen (Salzberg et al., 2008, BMC Genomics 9: 204), we found a 212-kb duplication. Later, this was confirmed to be real by PCR analyses using diagnostic primers at the edges of the duplication and flanking sequences, and by PacBio DNA sequencing (Adam Bogdanove, pers. common.) Saying this, that's one way to go.

Best wishes, Ralf

Hello Peter,

Thank you for reply. Yes, i have two different scaffolds having only 25kb regions with 99.75% similarity and also i have paired end reads.

With regards,

Rajesh

Hi Ralf,

Thank you for reply. I will see your publication.

With regards,

Rajesh

Rajesh, that sounds good. Then it may as well be a real duplication.

Still, agreeing with Ralf, theory crafting is helpful for deciding which next step to take, like reassembly with different parameters or subsets of the reads, etc. but the real thing is the wet-lab proof.

You may also need PCRs for proving and filing the gaps, anyways, so you can include some more for clarifying this question too.

Peter,

Well, its getting too confusing but I think only way out is to do some wet lab experiment.  Thanks for the valuable suggestions.

Regards,

Rajesh.

Rajesh, for me, your answer to Peter was ambiguous. Could you explain?

You have two different scaffolds S1 and S2, built from two different sets of distinct contigs (S1=C1,1, C1-2, ... S2=C2,1, C2,2, ...) separated by gaps. The contigs in a scaffold are ordered and oriented based on some cartographic constraints, for example, insert sizes of paired-end reads. You have observed two 25 Kbp regions in S1 and S2 that are almost identical, enough that you can't tell whether the differences might be an assembly artifact, or truly duplicated regions, or the results of RIPing, or whatever.

You didn't say, but I am guessing that this is a haploid genome of less than 40 Mbp, that the genes in the region of interest aren't ribosomal, that you have perhaps 5 genes in each 25 Kbp region, that each gene has 5-10 introns each, and that based on alignments at least these intron positions and sequences are also the same.

Could you tell us which of the following is true?

The more your assembler has isolated the two regions, the more you can suspect that it wasn't able to prove that the regions are truly duplicated. The more it has found different flanking regions, the more it looks like the regions are true duplicates.

Cheers, David

Dear David,

First of all, I am extremely thankful for your valuable response.

Well, yes we have two different scaffolds S1 and S2, built from two different sets of distinct contigs. You are right we have haploid genome and it's size is approx. 35Mbp. The putative duplicated regions have 10-12 number of genes with 1-6 introns, and based on alignments these intron positions and sequences are also the same.

In our case point 2 is true i.e. the two duplicated regions are part of two different contigs. Moreover, we were able to extract the 2kb upstream flanking regions from both of the contigs but we were not able to extract the downstream flanking regions because contigs ended there itself. However, we were not able to align or compare the isolated 2kb upstream flanking regions since the putative duplicated regions are reverse in orientation.

Looking forward for your suggestions.

Regrads,

Rajesh.