Nick Eriksson

... a mathematician, a different kind of mathematician, and a statistician

[Shai Carmi @ShaiCarmi]

Are you a clinician/researcher interested in the pros and cons of screening IVF embryos for polygenic diseases (PGT-P)?
Check out our new online risk reduction calculator.
https://t.co/Y36lLHtEbf

[Shai Carmi @ShaiCarmi]

An interesting new preprint on how to combine PRS and family history to improve disease risk prediction. By Alkes Price and colleagues.

Bottom line: adding disease status of family members as indicators to logistic regression works well.

Thread ->

1/14

https://t.co/in3e8A3XPz

[Nick Eriksson @nkeriks]

@samit_dasgupta there’s a turnout model described here, but no predictions available yet? https://t.co/QGi6VROolw

[Sumit Mukherjee @SMukherjee89]

I'll elaborate on this a little bit, since I feel it isn't just enough to say someone is my hero without explaining why. Daphne Koller (who I have never met in person) has had a disproportionately large impact on my life. Thread below (1)

[Nick Eriksson @nkeriks]

@Farzad_MD @atulbutte @KennethBaer @chiragjp And that map... when ~200 people in the sample are from 94022 (Los Altos & Los Altos Hills, probably the richest zip code in the county), it seems hard to write "targeting a representative sample of the county" with a straight face?

[Nick Eriksson @nkeriks]

@ewanbirney @matthewherper @skathire @jcbarret @SCSanderson Probably most important difference is disease frequency (e.g., figure 2 of https://t.co/5rMPLTzRgL) -- family history works well in practice only for quite common diseases (and this was with 7-year-ago "PRS")

[Nick Eriksson @nkeriks]

@GENES_PK @anshulkundaje @FiveThirtyEight Yeah, I’ve never written it out but it seems true. Re @anshulkundaje q about small differences in AUC, the most natural metric isn’t clear, but at least increasing one will increase others. It’s the nice part about small effect sizes in GWAS :)

[Nick Eriksson @nkeriks]

@anshulkundaje @GENES_PK @FiveThirtyEight Under the assumption of small effects sizes (for a fixed disease prevalence), the risk distribution is ~normal and there is just one parameter that all metrics can be derived from, so it doesn't really matter what metric we talk about. E.g., https://t.co/dt6V1M7IRF

[Sek Kathiresan MD @skathire]

Intense hunt by biotech (i.e. Calico) & billionaires to promote lifespan (healthy aging).

Here, a genetic study reveals the secret: avoid a heart attack!

Most gene variants identified for lifespan are just common alleles *protective* for heart attack.
https://t.co/K3KDtX420S

[23andMe Research @23andMeResearch]

Our Health R&D team is looking for a scientist to create novel health products. Background in epidemiology & machine learning a plus 📈👩‍🔬 #jobs #hiring #ASHGtrainee https://t.co/AYUAVxPduh

[Nick Eriksson @nkeriks]

Why-Because graph is what I'm calling decision trees from now on. #GRSwebinar

[Nick Eriksson @nkeriks]

"advantage to studying polygenic risk pre-GWAS: we worked with simulated data only" #grswebinar #notatypo

[Ariel Schwartz @ArielSSchwartz]

Fireside Chat with Daphne Koller (ICLR 2018). The best ICLR session IMO, discussing ML & computational biology, online free education, and women in science https://t.co/U3aBicgAyS

[dimenwarper @tsuname]

Deep learning, min hashing, assemblies oh my! If you're interested in such topics, join us @hexagonbio and start adding leaf nodes to this fungal tree!

[Karl Heilbron @karlheilbron]

My latest work w/ @23andMeResearch @predictPD on the Parkinson's disease phenome! Associations between PD and 840 survey-based phenotypes. Replicated 32 known associations and highlighted some potential risk factors (migraine, OCD, allergies, anemia). https://t.co/o6iejxgcBt

[ICERM @ICERM]

Congrats to Bernd Sturmfels, past ICERM scientific board member, who will receive the 2018 George David Birkhoff Prize in Applied Math from the AMS. #appliedmathematics #appliedalgebraicgeometry https://t.co/HO2ngfH8Qc

[23andMe Research @23andMeResearch]

New from our Team: GWAS & HLA region fine-mapping identify susceptibility loci for common infections https://t.co/rEvpvTsXds @NatureComms

[bioRxiv @biorxivpreprint]

Ingredients in Victoria's Secret Bombshell and Ivanka Trump eaux de parfums that repel mosquitoes https://t.co/vYU8stTNxS #bioRxiv

[Jordan Ellenberg @JSEllenberg]

Young mathematicians: how should you decide whether, and if so how, to move into private-sector data science? https://t.co/87w7uVWdT6