Could there be no SNPs in a homozygous genome assembly?

[Lisa0508]

Hi everyone, I am having a difficult time in dealing with the SNPs from a homozygous genome. Theoretically speaking, I suppose there should be no SNPs detected in the mitotic homozygous genome if no reads are misassembled in a de novo way. I used CLC and SOAPdenovo to assemble this homozygous genome, but still extracted more than 10,000 SNPs in the high coverage regions. Many frequencies were almost 50/50. Before performing the whole genome sequencing using this assumed homozygous individual, I used two microsatellite markers per chromosome to confirm its homozygosity. The genotyping results came out to be 100% homozygous at those loci. So whether it was due to the really variations on two chromosomes or due to the sequencing error, missassemly or something, I really have no idea. During assembly, there could be more or less mistakes, couldn't there? I think it's impossible to get no SNPs. But the boss wants me to prove its complete homzogyosity, now I am sunk. Could somebody show me some points?
Best,
Lisa

[adaptivegenome]

I think you misunderstand what a SNP is. But anyways first let me ask, is this a homozygous genome or haploid genome? This makes a difference on what you should expect.

[Lisa0508]

It's a homozygous(doubled-haploid) genome. Could you please explain to me?

[adaptivegenome]

If you sequence a haploid genome you would expect at any position in the genome, for all reads to be concordant (they should all agree). When they do not agree this is indicative of sequence or mapping error.

In a homozygous genome, depending the extent of inbred there can be regions of residual heterozygosity so this becomes an additional possibility if you find reads that do not agree. Two repeat markers on the chromosome are not sufficient to rule this out.

What you can do (and what lots of people do) is set a sliding window across each chromosome and mask out windows that have clusters of heterozygous calls.

Hope this helps.

[Lisa0508]

This truly helps. Could you please commend me some nice tools？Is the "VCFtool" fine for this purpose? But I am still wondering after masking the heterozygous calls out, how should I make the whole genome sequences complete? To fill those masked regions with "N"s?
Maybe I misunderstand your meaning again and make such strange questions. Really thanks a lot for your help.

[Lisa0508]

Even though we irradiated the sperm with high doses of UV, we still couldn't ensure that the genetic inactivation was complete. Some genetic fragments survived the UV irradiation and the offspring still got some genes from the paternal chromosomes.
Even what we used for sequencing was a mitotic one rather than a meiotic one, we still could not say it's 100% homozygous. Actually the so-called mito-gynogenetic individual just possesses a higher level of homozygosity than a normal one. But it can never be 100% homozygous. Am I thinking in a right way?

[adaptivegenome]

Yes. Basically there might be regions that are hard to assemble because of this. Broad has a new format [FastG] for dealing with assemblies that that have variations. Hopefully this will be of utility in the future.

[Lisa0508]

Thank you. Sorry for making so many questions. I read some papers about NGS assembly using the homozygous genomes. All claimed that they used the doubled-haploid ones as the materials. But none of them concerned the SNP problems. Search the internet, I found it hard to find one paper that discussed about analyzing or handling with the SNPs in the assembled sequences of the doubled-haploid genome. Which paper shoud be fine to refer to in dealing with the SNP problems? I badly need some reference for this.