Determining ploidy from fungal genomes
2
0
Entering edit mode
2.9 years ago
mss ▴ 40

Hi,

I am working on fungal isolates from deep-sea sediments. It is debated on whether these isolates are haploid or diploid. I was wondering if there was a way to determine if they were either haploid or diploid using some type of bioinformatics tool.

I did try assembling the genome using both SPAdes and diploid SPAdes. Diploid SPAdes seems to be the better assembly, but I do not think this means I can ultimately say that these fungi are diploid.

assembly haploid diploid fungi genome • 1.8k views
0
Entering edit mode

You can try kmercountexact.sh to test for ploidy (http://seqanswers.com/forums/showthread.php?t=64086 ).

1
Entering edit mode
2.9 years ago
toheitka ▴ 230

You might try to analyze the kmer distribution from your reads. For this, you might use

and count your kmers for a few different kmer lengths, e.g. 17mers as in the example.

Then, you would need to plot your data as in the figure below. X-axis: kmer depth, how often each kmer is counted, Y-axis: frequency, how many individual kmers are counted with this kmer depth.

The example is taken from the quinoa genome sequence by Zou et al. (2017). Quinoa is tetraploid. The graph is typical for a tetraploid organism as it has two peaks, a diploid and a tetraploid peak. A diploid organism would only have one peak,

You should check what it looks like for your species. Best luck!

1
Entering edit mode
2.9 years ago
Carambakaracho ★ 2.4k

Given that you have assemblies already and provided your strains are heterozygous wild strains (and not homozygous lab "wildtypes"), you can map the reads back to your assembly and predict variants on the assemblies. In haploids, you'd expect very few variants, in diploids many heterozygous mutations with an allele frequency distribution close to 50%. You can read this information directly in the vcf file from the variant caller and process it in Excel.

Using this, you can resolve up to tetraploidy, potentially even hexaploidy. I can't find the reference, though...

Probably not less work than toheitka's solution, but a different approach.