1/n Can LLMs perform scientific research? And, can they do so while enhancing key scientific values including transparency, traceability & verifiability? Check out our preprint on the 'data-to-paper' platform. w the amazing Tal Ifargan Lukas Hafner arxiv.org/abs/2404.17605

Tip for paper writing: when you have some unknown number or other blank to fill in, consistently use a symbol that is unlikely to show up elsewhere (I use ***). Then search for it at the end to avoid embarrassing inclusion of notes and missing values in the final manuscript.

We used multiplexed lineage tracing barcodes to analyze the emergence of resistance to mitotic kinesin inhibition in glioblastoma and to identify synergistic drug combinations:

Direct visualization of the kinetochore and centromere chromatin by cryo-ET. This is what maintains genome integrity at cell division. Bravo to PCGI co-Director Ben Black and Katie Kixmoeller on leading this tour de force.

biorxiv.org/content/10.110…

The shift towards translational science and AI seems to be a consequence of thinking that basic science problems are either solved or intractable, respectively. Perhaps that is true, or perhaps we are mortgaging our future. I honestly don’t know which.

one of the most important things I know about deep learning I learned from this paper: 'Pretraining Without Attention'

this what I found so surprising:
these people developed an architecture very different from Transformers called BiGS, spent months and months optimizing it and

When does development stop and decay start in a mouse or human? Is it the same for each tissue or different? How does one define when a tissue has reached peak maturity and starts to decline? I’d argue in humans that naive T cells are peak at 15-25 years and decline thereafter.

Michael Faraday grew up in a poor family.
He became an apprentice to a bookbinder to get access to books & self-taught himself.

Faraday's research on the magnetic field around a conductor carrying a direct current introduced the concept of the electromagnetic field in physics.

Great knows great. Yes, this is *the* David R. Liu. He had developed a neural network model of visual attention, which (unknown to him) was similar to something Hinton was working on. He was only 16 at the time. Did they meet? I don’t know yet.

Questions to ask: Is 'The automatic generation of sizeable synthetic transcriptomics datasets, faithful to the data distribution, is a sound way to address the shortage of [RNA-seq] data' an assertion that needs to be proven, or one that can be taken as axiomatic in the intro?

Automated methods have many benefits, but one that is often touted that I disagree with is that they are “unbiased”. I see many people who seemingly prefer a clearly flawed but automated pipeline to manual analysis. That is troubling.

Wait are the units of time here months? Or weeks, days, or hours? I’ve seen (successful) examples of all of the above. :)

Happy to share that our work on how small set of motifs and rules explain most human promoters is published in Science Magazine I am very grateful for the chance to be part of a team of amazing scientists: Donghong Cai, Chenlai Shi,Jian Xu,Jian Zhou science.org/doi/10.1126/sc…

The d2l authors should spend some time with microscopy data to disabuse themselves of this notion :)

does fusing antibodies to AAV capsids actually work? what is the best published data on this?

Knowing when an email should be a phone call is my last bit of Gen X comparative advantage.

I don’t know if I’m more confused about Fisher Scientific selling jeans, or about this being the top search result for a typo in paraformaldehyde.

I feel like I have just quantified how much my brain has shrunk in the last 20 years.

Excited to share new results from our 4DN collaboration between the Gilbert, Ma, van Steensel, and Belmont research groups. Joint analysis, led by our trainees, compared chromosome positioning relative to multiple nuclear locales across 4 cell types
doi.org/10.1101/2024.0…

WHAT % OF YOUR TARGET BIOMOLECULES ARE YOU REALLY IMAGING?

The *labeling efficiency* of your experiment is extremely important for image quality and quantitative results in super-resolution!

Super excited about our new article in Nature Methods! 🤩

nature.com/articles/s4159…