Re: [svtoolkit-help] Running GenomeSTRiP on non-human organism

[Bob H. <han...@br...>]

Hi, Ana,

The documentation is reflective of the fact that I haven't tried 
anything like this and I can't promise that there won't be some gotchas.
But from my brief reading of the ZW system (on Wikipedia), I think you 
could try using a custom ploidy map (details below) and see how far you get.
I would be interested to learn where things do and don't work.

First, you should determine if there are pseudo-autosomal regions on Z 
or W, like in human, and in particular how your reference sequence 
models these.
This has as much to do with how the reference is built and how the 
alignments are done as it does the acutal biology.

The hg19 ploidy map we use looks like this:
X 2699521 154931043 F 2
X 2699521 154931043 M 1
Y 1 59373566 F 0
Y 1 59373566 M 1
* * * * 2

Where we are specifically indicating the non-PAR region on X (as opposed 
to all of X).
Note that the human reference includes the PARs on X but masks out the 
PARs on Y.

If you have no PARs, then you might try something like:
Z 1 nnn M 2
Z 1 nnn F 1
W 1 nnn M 0
W 1 nnn F 1
* * * * 2

Basically, the ploidy map is giving the expected ploidy for a sample 
with a certain gender in each region of the genome.
You can specify as many regions as you need.  The last line is a 
wild-card saying the rest of the genome is diploid.

The preprocessing code is hard-wired to do an X/Y sex chromosome 
analysis and automatically call sample gender.
This may fall over or it may just not produce any output (which would be 
the more desirable).  Please let me know if it falls over.
Since there won't be any automated gender calls, you will have to supply 
the gender of each of your samples.
The preprocessing code is using read depth across X and Y (with a custom 
mask that helps clean up the signal) and then calling gender from the 
gross copy-number estimates.
You could also do something similar, either to call gender or to QC your 
gender assignments.

Hope this helps,
-Bob

On 10/13/15 10:56 AM, Ana Pinharanda wrote:
> Dear Bob,
>
> I have installed and run the test data for svtoolkit_2.00.1602 with no 
> errors.
>
> I work with a non-human female heterogametic (ZW) system and while 
> reading the documentation under
> _http://www.broadinstitute.org/software/genomestrip/node_ReferenceMetadata.html_
> on how to built a custom reference metadata bundle, I came across the 
> following statement under the /reference./ploidymap.txt section
> "Currently, only organisms with a human-like XY sex determination 
> system are fully supported."
>
>
> My question is: what does this actually mean?
>
> Does it mean that they are not supported just simply because there are 
> no examples of any of them under your reference metadata bundles?
> Or does it mean that for my runs to work I need to swap the Z for a X 
> and the W for a Y when I make the ploidymap.txt?
> Or does it mean that beyond swapping the letter (Z->X & W->Y) I also 
> have to label males as females and females as males so that the 
> predictions of a male heterogametic system match up with what I am 
> feeding into it?
>
> Finally, what would your predictions be if I dropped any W linked 
> regions altogether from the analysis? Would a Z0 system be allowed? If 
> so, how would I make the ploidymap be able to reflect this correctly?
>
> Thank you in advance for the time spent,
>
> All the best,
>
> Ana
>
>
>
>
> ------------------------------------------------------------------------------
>
>
> _______________________________________________
> svtoolkit-help mailing list
> svt...@li...
> https://lists.sourceforge.net/lists/listinfo/svtoolkit-help