Extracting sex chromosomes from sequenced genomic DNA reads or assembled genome

[NGS_New_User]

Hi everyone.

I have two sets of data; male and female (Illumina sequenced, PE reads, ~300 million each, 100bp) of a non-model organism (insect). Since it has no reference genome I performed a de novo assembly on both sexes.

I would like to extract the sex chromosomes from each sex.
For the Y chromosome, I mapped all the male reads to the female draft genome and the unmapped reads that remained I grouped them as the 'Chr Y' reads. I then performed an assembly of the unmapped reads to get the 'ChrY' contigs. Does this sound like a reasonable way to tackle this?

I am a little stranded on how to get the 'Chr X' reads. Does anyone have any idea of how I can do this, or any suggestions? It will be highly appreciated.
Thanks.

[Linnea]

Hi,

For the ChrY I would actually suggest that you assemble the male reads instead, then map the female reads and define the ones with zero female read coverage as Y scaffolds. Doing the other way around, you would surely get the Y reads but also any contamination or "crap"-reads from the sequencing.

We tried both of above in a non-model organism (bird) and it worked much better with the assembly-first approach. (Our paper is here: http://www.nature.com/ncomms/2015/15...comms8330.html)

The method is partly taken from here: Hall, A. B. et al. Insights into the preservation of the homomorphic sex-determining chromosome of Aedes aegypti from the discovery of a male-biased gene tightly linked to the M-locus. Genome Biol. Evol. 6, 179–191 (2014).
And they have a script here: http://tu08.fralin.vt.edu/software/CQcalculate

To get the ChrX is trickier, but one thing you can try is to map both female reads and male reads to the female assembly and compare the coverage quota (of course normalized if you have different amounts of female and male data). All chromosomes should in theory have similar (normalized) male and female coverage except for the X-chromosome, where the M:F ratio should be 0.5. Because of sequencing bias of course not all scaffolds will have the same coverage, so it's probably best to use this method in combination with some other method, like blasting the scaffolds to a related species with a defined X-chromosome. For our organism it worked well when coverage was very high (>100X).

Good luck!
Linnéa

[NGS_New_User]

Hi Linnea,

Thanks for the suggestion and the article links. This should be extremely helpful, I am grateful! I had thought of the 'crap reads' in ChrY but I wasn't sure yet of how to handle them. I will try your suggestion and compare the two.

I am currently trying to get the ChrX using the same approach that you have suggested. However since we have quite a fragmented assembly (~200k contigs) I am getting a lot of contigs with varying M:F ratio of between ~0.45-0.55. Our closest related species would be Musca domestica but since it is not yet completely annotated, I guess I can try using the ChrX from Drosophila melanogaster.

Thanks!