Missing call for ZNF713

[Paranchai87]

Hi there,

there is a polymorphic deletion of variable length near the ZNF713 repeat. If too long, trgt will not call a variant for this gene.
Is there a logic that you use 250 bp flanking as default or is this arbitrary? One of my samples has a nearby polymorphic deletion of 91bp, for which a repeat was not called (with this default 250bp flanking). I then tried to rerun with a series of flanking numbers (50,100,150,200,240,245), and found that all called the repeat variant. So, in this case, the cutoff can be 245bp. I have another sample with 135bp polymorphic deletion and am trying to find the cutoff, although I don't know how long this polymrphism can be. I looked quickly through other clinically relevant 71 loci and found no adjecent polymorphic large deletion. So, I was wodering if it is better/safer to set this default flanking lower (e.g. 200bp) for all loci. What would you recommend? If you want to check, the HG002 sample also has this polymorphic deletion nearby.

I am looking forward to hearing from you soon.
Best, Paranchai

[egor-dolzhenko]

Hello Paranchai,

Thank you for the question. As you observed, selecting optimal flank length can be difficult because it can differ by repeat and by sample. A solution that worked well for us is to extend the boundaries of a repeat region to encompass all surrounding variation. This led us to develop some basic tools for variation cluster (VC) analysis described in this paper:

https://www.biorxiv.org/content/10.1101/2024.10.04.615514v2

In particular, VC analysis extends the boundaries of the ZNF713 repeat from chr7:55887600-55887639 to chr7:55887600-55887904. Could you please confirm if this region encompasses the deletion you are describing?

Relatedly, you can explore repeats and VCs containing them in a new browser created by @bw2.

https://trexplorer.broadinstitute.org/

Best wishes,
Egor

[Paranchai87]

Hi Egor

Many thanks for your answers. However, I think you could misunderstand what I meant.
It is not about the region the repeat is located, but the flanking sequence, ie number of bases...right? Which is --flank-len <FLANK_LEN> Minimum length of the flanking sequence [default: 250]. I tried to run with several numbers and found that --flank-len 190 still gave calls for all 73 Loci, including the ZNF713 repeat.

Although I think it is not relevant, the answer to your question is yes, that chr7:55887600-55887904 encompasses the deletion polymorphism I mentioned.

Or what do you mean?

Best wishes
Paranchai

[egor-dolzhenko]

Thank you, Paranchai. Using 190 bp flanks for all 73 repeats you are studying seems reasonable. However, to be on the safe side, I would suggest performing additional benchmarks to confirm that changes to flank length do not result in significant changes to genotypes of these repeats in your samples.

In general, optimizing flank length for all repeats across the genome is a tricky problem. This is because for some repeats 250bp flanks work better than 200bp flanks while other repeats (as you observed) are easier to resolve with shorter flanks.

This is the reason I mentioned our work on adjusting repeat boundaries. Adjusting repeat boundaries allows us to avoid optimizing flank lengths (and other analysis parameters) for many otherwise problematic repeats. For example, since the extended ZNF713 region (chr7:55887600-55887904) encompasses the deletion you are describing, I expect that this updated definition would require no flank length adjustment.

I hope this is helpful. Please let me know if you have any further comments.

Best wishes,
Egor

[Paranchai87]

Dear Egoz

thank you so much for your clarification. Now, I understand. I just extended the locus definition for that gene, but keep the default flanking to avoid possible misscalling for other loci, and it worked.

Have a nice day.
Best
Paranchai