Indel treating

[nguyenvinh dao]

Hi!

I am working with about 5000 human genomes with 0.2x. Apart from these samples, I have 14 samples with wgs (higher depth).

I intend to use these 14 samples to construct the gen.txt file. Then, I subsampled these 14 wgs and add to the pool for analysis.

I used position of SNV reported by KHV dataset (https://genomes.vn/) to construct the pos file. Although I only focus on SNV only, in many situation, the corresponding positions among 14wgs is of indel. How do I treat these indel? Remove it from the analysis?  Or I should treat it as SNP with alternative value?

In many situations, there are more than one alternatives. What should I put in the fourth columns (column for alternative)?

Thanks for your help!

Cheers, Vinh

[Robbie Davies]

Hi,

Since this is human data, I recommend you consider QUILT, using the 1000 Genomes reference panel, which includes some KHV samples. Great you have 14 high coverage samples.

https://github.com/rwdavies/QUILT

If you have time, you could also run STITCH and see how they compare, given you have so many samples.

For both QUILT and STITCH, unfortunately they only take bi-allelic SNPs. So you'd have to remove non-bi-allelic SNPs from your analysis. I believe it is possible to re-impute afterwards using e.g. Beagle or another software if you're very keen on the indels. Apologies that neither QUILT or STITCH support indels. Unfortunately it's not feasible for me to add indel support in the foreseeable future.

Best,

Robbie

--
You received this message because you are subscribed to the Google Groups "STITCH imputation" group.
To unsubscribe from this group and stop receiving emails from it, send an email to stitch-imputat...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/stitch-imputation/33fbb1bc-c58c-44d9-8c9b-c0d516d123bcn%40googlegroups.com.