Coverage Experiment To Indicate Polyploidy.

Question

[Fabian Bull]

I am searching for hints of polyploidy in two species:

My hypothesis is that species A is diploid and B is tetraploid.

Because I have different sequencing depth in the 2 species I extracted reads in the ration 1:20 ( ratio of the genome sizes) to get similar expected coverages.

These two sets of reads are de-novo assembled independently and the coverage of the created contigs is analyzed. A histogram of the coverages is given below:

You can see that the peak species B is approx. at the double of peak A. The different tail shape can be explained by a different structure in repetitive elements.

Questions:

1. Do you see any flaws in this experiment?

2. Could there be other reasons for the shifted peak?

3. Can you imagine other experiments which could indicate polyploidy?

Edit: If you agree with my throught-process you should also add a comment/answer.

[Istvan Albert]

I think I understand what you are after - basically more DNA coming from one experiment indicates that it has more copies - but how did this really work - what does it mean to have extracted reads in the ration of 1:20 were the samples sequenced separately or together

[Fabian Bull]

They were sequenced and assembled separately. My idea was that maybe the duplicated regions collapse in the assembly and therefore are mapped twice as often as the other regions.

Answer 1

[Chris Miller]

What about plotting the allele frequencies for all heterozygous SNP sites? Assuming that there's a little bit of CN in these genomes, you'd expect to see the following:

• diploid organism: big peak at 50% (neutral, CN2) , smaller peak at 33% and 66% (+1 copy CN3)
• tetraploid organism: peaks at 25%, 50%, 75% (neutral, CN4). Possibly smaller peaks at 33/66% (-1 copy, CN3) and smaller peaks at 20,40,60,80 (+1 copy, CN5)

This approach depends on having deep enough coverage to resolve those peaks. If not, you'll just get a big smear.

[Fabian Bull]

Could you explain what you mean by CN? I assume copy number? CN2 are snps which occur in 2 variations? Sorry but I am not really into the snp terminology.

[Chris Miller]

yes, CN2 = diploid (two copies of that gene). If it's a heterozygous site, say C/T, and you count the number of reads containing each of those alleles, then roughly 50% should be C and 50% should be T. If it's a region that has three copies, again with a het snp C/T, then your options are 33% C and 66% T, or 66% C and 33% T. There wil be some variation around this number due to sampling error, but if you look at lots of sites, you'll see peaks emerge.