Computer scientist researching AI for population genetics, ancestry, and genetic genealogy. I work at Apple and prev. at Amazon, Dolby, Microsoft, Ancestry.
San Francisco, CA
Joined January 2014
- Tweets73
- Following883
- Followers1.4K
- Likes1.5K
Geleta retweeted
A fantastic turnout for HPRC's workshop at #ASHG2026!
Thank you to everyone who joined us to learn more about the Human Pangenome Release 3 and its data, tools, applications, and path forward.
Geleta retweeted
New paper from @padillaccc looks at how social norms such as inbreeding taboos / exogamy, mobility, and "mating distances" affect homozygosity in modern hunter gatherers from Central Africa.
A very cool synthesis of ethnography and genetics
Read it here:
science.org/doi/10.1126/scia…
Next week I will be at Montreal, Canada🍁 for the @GeneticsSociety meeting. I will be presenting PCLAI, our new method for continuous local ancestry inference, at my poster on Oct 21, 2:30-4:30pm EDT, and speaking at the HPRC workshop on Oct 24. biorxiv.org/content/10.64898…
Geleta retweeted
Here are the key figures for that. We think an ultra-voyaging culture originated out of remote French Polynesia (Mangareva) and settled both Easter Island (cool far south Pacific) and Hawaii (only Polynesian island in the northern hemisphere) with related individuals settling both far-flung places.
Geleta retweeted
Just the Editor-in-Chief of Nature advising scientists not to share their data and findings with external AI model companies.
Geleta retweeted
This is a super cool approach to measure the structure of genetic variation
Fst is a key genetic differentiation metric, but it emphasizes common variants over rare.
In @ieeeICASSP out now @geletavc introduces Fq, which can be tuned across the allele freq spectrum.
Fst & Shannon differentiation fall out of it as special cases.
arxiv.org/abs/2511.03063
Geleta retweeted
For different q, Fq weights rare variants more (or less), so can focus more on them. For q=2, Fq is just Fst, which weights common more & for q=1 it is Shannon differentiation, up-weighting rare more than Fst. Fq gives a theoretical structure for connecting these & for weighting.
Geleta retweeted
ICLR 2027 has received more submissions than all previous years (2013–2026) combined
Geleta retweeted
Super cool stuff. I’ve had a strong intuition about the connection between FST and entropy / information for a long time, but didn’t know about Tsallis entropy. Reading this makes the connection even clearer and much cleaner
Fst is a key genetic differentiation metric, but it emphasizes common variants over rare.
In @ieeeICASSP out now @geletavc introduces Fq, which can be tuned across the allele freq spectrum.
Fst & Shannon differentiation fall out of it as special cases.
arxiv.org/abs/2511.03063
Geleta retweeted
TMLR has faced a deluge of submissions, necessitating stricter desk rejection policies due to limited reviewer capacity
Co-EiC Nihar Shah reached out to authors of 10 papers slated for desk reject. Could they answer questions about their *own* submission?
medium.com/@TmlrOrg/asking-a…
Geleta retweeted
Her Tsallis-order q fixation index, Fq, measures the fraction of Tsallis q-entropy lost within subpopulations relative to the pooled population, placing familiar measures inside a single entropy family.
• q=2 recovers the classical heterozygosity-based Fst
• q=1 gives Shannon differentiation (i.e., normalized mutual information), the absolute q=1 form is mutual information between allele identity and population label
Instead of looking at only one statistic, we can now examine a spectrum across q (and its slope).
Within 1≤q≤2, lower q gives more weight to rare variants; higher q increasingly emphasizes common variants. This can expose differentiation patterns that a single Fst value may obscure.
Geleta retweeted
Fst is a key genetic differentiation metric, but it emphasizes common variants over rare.
In @ieeeICASSP out now @geletavc introduces Fq, which can be tuned across the allele freq spectrum.
Fst & Shannon differentiation fall out of it as special cases.
arxiv.org/abs/2511.03063
Geleta retweeted
If anyone here is at EMBL's "Reconstructing the human past" in Heidelberg (near where Heidelbergensis fossils were found, fun fact) I'm going to be presenting this as a poster, and would love to chat about my preprint (or anything else ancient DNA related 😁) #EESHuman
Well, what did you expect? A stereotypical test (of course) will output stereotypical results. Correlation ~0% Sun's sign & psychometrics checks out. Astrology is an occult practice, most of it still buried in Chaldean texts. I'd be surprised if LLMs performed well.
We cross-referenced the psychometric personality data of ~ 1 million people with their zodiac astrological signs:
1) There is a small but statistically significant effect of zodiac sign on personality controlling for demographic and attitudinal variables
2) These observed personality correlations correspond strongly with LLM-generated "stereotypes" of astrological associations with personality
3) This association of observed astrological effects with astrological "stereotypes" is mediated strongly in auxiliary surveys by how frequently people read horoscopes and generally subscribe to astrology
4) Zodiac effects among people who never read horoscopes are a quite precisely estimated zero, while effects among people who read horoscopes everyday are on the order of 0.2 sds - on par with or larger than most psychometric associations found in social science
Geleta retweeted
Staff Scientist position at University of Copenhagen. Come join our work on ancient environmental DNA! candidate.hr-manager.net/App…
Geleta retweeted
ok but what does my sexual orientation have to do with submitting a paper to science
Geleta retweeted
ARGformer is out in Bioinformatics!
doi.org/10.1093/bioinformati…
Just as PCA produces a low-dimensional representation of population structure from genotypes, ARGformer can do this now for ARGs...a representation that comes now directly from inferred genealogical topologies & at a haplotype level across each individual's genome.
We do this by turning local genealogical paths into tokens & learning dense embeddings, similar to language embeddings, to visualize, cluster & query.
Code
github.com/AI-sandbox/ARGfor…
Team
@d_bonet, Cole Shanks, @marcal_cc, @jordiatjhu
