Karsten Sieber

Human genetics & genomics, with a bit of data science. Tweets represent personal views only.

[Itai Yanai @ItaiYanai]

Novelty needs time. For the first 3 years after publication, a highly novel paper is less likely to be cited than a relatively non-novel paper. But after 15 years, they are 60% more likely to be highly cited. Don't let the short-term view discourage risky research!

[Sasha Gusev @SashaGusevPosts]

I highly recommend anyone working with single-cell data read this and take these issues seriously. These tools are under active development, which is terrific and to be applauded, but has also created a very fractured methods ecosystem. 1/

[Daniel MacArthur @dgmacarthur]

Excellent policy shift from the Gates Foundation, mandating preprints and no longer supporting OA fees. This will ensure research products are accessible without propping up the fundamentally broken current publishing model. Hope other funders follow! https://t.co/rQN8XWSNZS

[Karsten Sieber @siebzscience]

@cgonzagaj Following that logic, you must also be opposed to the UKBiobank, FinnGen, & All of Us projects?

[Lior Pachter @lpachter]

It's time to stop making t-SNE & UMAP plots. In a new preprint w/ Tara Chari we show that while they display some correlation with the underlying high-dimension data, they don't preserve local or global structure & are misleading. They're also arbitrary.🧵https://t.co/XkAOTKlOcs

[Karsten Sieber @siebzscience]

@DrMikeWiser put them together on a bookshelf(s?) of shame

[Karsten Sieber @siebzscience]

@ChBurkhart I *assume* the default behavior for implicit is to use lazy eval = TRUE while explicit (`future()`) has the lazy eval default as FALSE.

[Karsten Sieber @siebzscience]

@dagarfield @lal_avantika @doctorveera Don't you think that indication expansion for an approved drug is probably more important?

[Karsten Sieber @siebzscience]

@ChBurkhart My biggest barrier was getting my brain from `df %>% map(...)` to `df %>% mutate(foo = map(...))`

[Karsten Sieber @siebzscience]

@DanielDSeaton Great work Daniel. Wish I was at ASHG to stop by and chat. Hopefully next year!

[Michael 英泉 Eisen @mbeisen]

The underlying premise is extremely simple: authors publish their own work as preprints, and journals publicly review these already published works.

That's it. That's the system.

[Karsten Sieber @siebzscience]

@mbeisen I think that a major component that is missed in this question is that GWAS can influence therapies even if the original drug wasn't derived from a GWAS hypothesis. For example, GWAS can support new indications for existing molecules.

[Karsten Sieber @siebzscience]

@MarigortaLab @dccc_phd @zkutalik @ElePorcu @Luke0connor LoF mutations would be the only thing close to FC. The *vast* majority of “TWAS” like changes are very small relative to DEG, but each analysis has different goals. “TWAS” like analyses on FC scale are useful for understanding therapeutic intervention.

[Karsten Sieber @siebzscience]

@nilshomer Also known as natural selection 😉

[Karsten Sieber @siebzscience]

@cristenw 100% agree. This also happens to men with paternity leave. Is there a word to describe this effect for all parents?

[Livia Puljak @liviapuljak]

Our new study shows that data availability statements are not very useful; 1670 (93%) authors who indicated that data are available on request either did not respond or declined to share their data with us. Journal of Clinical Epidemiology: https://t.co/4IT2Dgphl4

[Karsten Sieber @siebzscience]

@PythonMaps @DrNishanShaik Countless domains have a language of preference which would be the top priority to learn. Advice to all careers to learn one language ludicrous. Your bias is blinding you.

[Karsten Sieber @siebzscience]

@PythonMaps @DrNishanShaik 🙄

[Karsten Sieber @siebzscience]

@jimhester_ Interested in a solution for this too ...

[Karsten Sieber @siebzscience]

@patsule @Eric_Fauman @ksuhre Agreed, -log10 *needs* to become standard