shaicarmi.bsky.social
Associate professor at the Hebrew University of Jerusalem.
Statistical, population, and medical genetics; preimplantation genetic testing. Views my own.
http://scarmilab.org
168 posts
689 followers
176 following
Regular Contributor
Active Commenter
comment in response to
post
Oh wow, I have the same issue with Ghost 14 (otherwise incredible shoes). Nothing keeps them tied. Good to know it's not me...
comment in response to
post
Another thing I find really weird is years-long enthusiasm regarding "iterated embryo selection". In more standard language, it's repeated selfing, each selecting for the best polygenic scores in each iteration.
Only that you end up with an entirely homozygous genome, which is probably not so great.
comment in response to
post
I mean, I think it's almost a no-brainer that in 20 years embryos will get their PCSK9 gene edited for reduced LDLC. But it's a long way before editong hundreds of variants for "super babies" with an IQ of 170 and a lifespan of 130 years.
comment in response to
post
The type of thinking is challenging to debate. The big picture is reasonable and promising. It's clearly well thought out. But it's the fine technical details that they get wrong.
In that article - the fact that causal variants are unknown and not coming anytime soon despite billions in investment.
comment in response to
post
Our study sheds light the fascinating world of ploidy aberrations. These embryos do not survive to birth. But their genomes reveal the various unexpected ways in which meiosis can go wrong.
Identifying and understanding these errors will contribute to explaining meiosis and infertility.
9/9
comment in response to
post
We found a maternal age effect, whereby the risk of triploidy increases by 76% between ages 30 and 40.
Finally, ploidy aberrations tends to cluster in families: six couples had three affected embryos (four with female age<35), unexpected by chance, suggesting a genetic basis.
8/9
comment in response to
post
Using another dataset of 74k embryos, we found 41 with "genome-wide uniparental isodisomy". These diploid embryos have two copies of the maternal genome and no paternal genome.
Naturally, these embryos have high levels of homozygosity.
How the paternal genome disappeared is unknown.
7/9
comment in response to
post
Remarkably, we found seven embryos that did not have a single crossover!
These embryos had both maternal chrs genome-wide.
(Except chrs with an additional aneuploidy; see figure for SNP array validation of one such embryo.)
6/9
comment in response to
post
We identified crossovers in triploid embryos based on switches between having both maternal homologs and just one.
We found less crossovers than expected. Further, embryos with less crossovers had more additional aneuploidies in specific chrs (on top of the triploidy).
5/9
comment in response to
post
We inferred the source of the error using two approaches:
1) Sex chr composition (XXX/XXY/XYY etc)
2) 55 embryos with parental data
Findings:
1) Almost all triploid errors are maternal, with ~1/3 from meiosis I and ~2/3 from meiosis II
2) Almost all haploid errors are paternal
4/9
comment in response to
post
We started with 96,660 embryos that underwent PGT for aneuploidy using Juno's targeted sequencing method.
All were normally fertilized ("2PN") and developed to the blastocyst stage. We found 882 triploid embryos (an extra copy of all/most chrs) and 181 haploid (missing copy of all chrs).
3/9
comment in response to
post
The study was led by Antonio Capalbo and his team at Juno Genetics, particularly Ludovica Picchetta and Christian Ottolini. It was co-led by
@eva-hoffmann.bsky.social
2/9
comment in response to
post
Our study sheds light the fascinating world of ploidy aberrations. These embryos do not survive to birth. But their genomes reveal the various unexpected ways in which meiosis can go wrong.
Identifying and understanding these errors will contribute to explaining meiosis and infertility.
9/9
comment in response to
post
We found a maternal age effect, whereby the risk of triploidy increases by 76% between ages 30 and 40.
Finally, ploidy aberrations tends to cluster in families: six couples had three affected embryos (four with female age<35), unexpected by chance, suggesting a genetic basis.
8/9
comment in response to
post
Using another dataset of 74k embryos, we found 41 with "genome-wide uniparental isodisomy". These diploid embryos have two copies of the maternal genome and no paternal genome.
Naturally, these embryos have high levels of homozygosity.
How the paternal genome disappeared is unknown.
7/9
comment in response to
post
Remarkably, we found seven embryos that did not have a single crossover!
These embryos had both maternal chrs genome-wide.
(Except chrs with an additional aneuploidy; see figure for SNP array validation of one such embryo.)
6/9
comment in response to
post
We identified crossovers in triploid embryos based on switches between having both maternal homologs and just one.
We found less crossovers than expected. Further, embryos with less crossovers had more additional aneuploidies in specific chrs (on top of the triploidy).
5/9
comment in response to
post
We inferred the source of the error using two approaches:
1) Sex chr composition (XXX/XXY/XYY etc)
2) 55 embryos with parental data
Findings:
1) Almost all triploid errors are maternal, with ~1/3 from meiosis I and ~2/3 from meiosis II
2) Almost all haploid errors are paternal
4/9
comment in response to
post
We started with 96,660 embryos that underwent PGT for aneuploidy using Juno's targeted sequencing method.
All were normally fertilized ("2PN") and developed to the blastocyst stage. We found 882 triploid embryos (an extra copy of all/most chrs) and 181 haploid (missing copy of all chrs).
3/9
comment in response to
post
The study was led by Antonio Capalbo and his team at Juno Genetics, particularly Ludovica Picchetta and Christian Ottolini. It was co-led by @eva-hoffmann.bsky.social.
2/9
comment in response to
post
Yes, I agree. But perhaps the effect size won't overflow (as opposed to the p value).
comment in response to
post
Interesting example! When working with p-values so small, wouldn't it be better to prioritize variants by effect size rather than p-value?
comment in response to
post
For example, it depends on the definition of the disease, and on which ages, environments, and sub-populations were used to estimate the risk. We have to make decisions on what we're estimating, even when sample sizes are very large. 3/3
comment in response to
post
Re. your question, let me try to refine the argument. For LDLC, there is an actual number of molecules in the sample. We can (hypothetically) count them and always reach the same answer. A genetic risk is a statistical construct, which depends on assumptions. 2/3
comment in response to
post
I'll clarify that I'm not against adding PRS to risk models, despite the variability. Quite the opposite. And I don't think the variability is a statistical problem (I agree with @mikeinouye.bsky.social).
(But it's certainly a serious clinical/practical problem.) 1/3
comment in response to
post
Interesting, I thought measuring LDL-C was a solved problem. (Or are things better now?)
Anyway, I think the problem with PRS is that (in contrast to LDL-C) there's no "true" risk that just needs to be measured better. Variability is not only technical. It's also which GWAS/method is used etc.
comment in response to
post
Police agencies probably need to show they are making efforts, hence buying these meaningless portraits. However, it is not only unhelpful, but it could also be harmful, given that an "ancestry-average" portrait implicates entire communities, usually minorities.
comment in response to
post
Most likely, their tool is just inferring broad ancestry and generating the average face for the detected ancestry.
Face prediction beyond "ancestry-average" is not currently feasible (and may never reach forensically-relevant accuracy).
comment in response to
post
They also market a tool for generating a portrait based on genotypes. They already sold it to hundreds of police agencies.
Their tool is based on a 2012 research that was never published. They never shared any information on how their prediction works and how (if at all) it was validated.
comment in response to
post
I think there two main differences compared to LDLC:
1) The PRS cannot be modified
2) The PRS depends on the GWAS/algorithm/etc, so can vary across labs (and it's generally more difficult to explain)
I think it justifies a higher bar.
comment in response to
post
Nice review!
Please be aware of these additional genomic projects/resources that were developed here:
medicine.ekmd.huji.ac.il/en/imgd/Page...
psifas.org.il/en/%d7%93%d7...
Partly also this: link.springer.com/article/10.1...
comment in response to
post
One more tool for compressing variant calls:
academic.oup.com/bioinformati...
comment in response to
post
Here you go:
bsky.app/profile/natc...
comment in response to
post
Thread 🧵 about our new paper on multiple testing in the IBD rate-based selection scan. Having a lot of recent relatedness at a gene (high IBD rate) can be evidence of strong selection, but the IBD rates are correlated. We want to correct for that.
www.biorxiv.org/content/10.1...
comment in response to
post
See also www.biorxiv.org/content/10.1...