Ha! Thanks, I'll file right under "Even reviewer 3 was speechless"

Thank you to other co-authors @carolinefwright.bsky.social, @mnweedon.bsky.social, @timfrayling.bsky.social, and @drarwood.bsky.social, @nihrexeterbrc.bsky.social, biobanks @ukbiobank.bsky.social and All of Us, and all of the participants of the studies

A massive, massive thank you to @hiwwright.bsky.social, @rnbeaumont.bsky.social, @drghawkes.bsky.social who all really drove this project to completion. Without them I would still be twiddling my thumbs shouting to the clouds about QC (I still will, but now you can read about it too!)

Using our DNANexus applet it is fast and not too expensive to QC and convert the entire UK Biobank WGS files to pgens! github.com/chundruv/ukb...

And for All of Us v8, we will provide you with the QC’ed pgen files on publication in a public workspace available to registered users.

“Ok fine, but what about the X chromosome, you always forget that”

This time we didn’t ignore the X chromosome! We show that you should pay special attention to non-pseudoautosomal X chromosome where QC should be more lenient for haploid males.

“But Kartik, how do we know the genotypes are wrong?”

Trios! Both cohorts have ~1k parent-offspring trios that were recruited incidentally.

Applying genotype-level QC reduces Mendelian errors by ~60-80% (even in All of Us where they already did genotype-level QC on hom-refs!)

“Bah humbug! How bad could it be?”

After genotype-level QC and a 10% missingness cut-off, we remove ~100 million (~9%) variants!

Most genotypes removed are homozygote reference (which were filtered in All of Us already)

We caution that the released data in UK Biobank and All of Us is not as clean as you may believe!

Here, we show how we determine data quality in WGS data, provide a really fast way of doing so on biobank data, and we will release QC’ed plink files for All of Us upon publication

I missed Hyung Chul’s poster! I’ll have to go see it later today