Minjune Song
@minjunesh
i trusted papers once
Short Sleep Variants Do Not Replicate In March Kristjan Moore from deCODE genetics emailed me urging me to stop working on short sleep. I read paper in question since Jan, but was still hoping/coping that maybe NPSR1 variant (not included in the paper) would replicate.
Wow, so it begins. In protein folding the CASP challenge set the grounds for domain outsiders to ignore domain-expert priors and focus on ML, birthing AlphaFold. Beginning of AlphaFold moment for cell state prediction?
Dropped the Virtual Cell Challenge Primer on HF. We are shipping transformers support for STATE (the SOTA model for predicting perturbation response) very soon!
Modified 3D printer pressure field mapping: 50µm resolution, $300 vs Onda's AIMS III 3-axis motorized scanner: 5µm, $90k ondacorp.com/scanning-tank/
we've wasted a few decades of highly intelligent human labor in the last period of history where it was abundant, and created literature that has negative utility for training scientific AI since it is untrustworthy as ground truth. Much will have to be redone from scratch.
Short Sleep Variants Do Not Replicate In March Kristjan Moore from deCODE genetics emailed me urging me to stop working on short sleep. I read paper in question since Jan, but was still hoping/coping that maybe NPSR1 variant (not included in the paper) would replicate.
PGS and GWAS replicate in studies with >>100k cases and controls, often in validation cohorts from different continents, born in different decades or even different generations, etc. Candidate gene studies, not so much. Most biomedical studies are powered more like the latter…
Short Sleep Variants Do Not Replicate In March Kristjan Moore from deCODE genetics emailed me urging me to stop working on short sleep. I read paper in question since Jan, but was still hoping/coping that maybe NPSR1 variant (not included in the paper) would replicate.
The concept of using Optimus to build on Mars is great, but I think that same concept can be used before Mars to build in previously inaccessible areas for humanity, or at least to increase the rate at which we can build in those places. Imagine a crew of robots building in the…
Who wants to do this with me? Got some cloud credits for one hail mary training run. virtualcellchallenge.org
To appreciate the enormity of what is happening you have to track multiple sectors - it's happening in almost every area of science and technology. docs.google.com/presentation/d…
Total arXiv victory in Grok4 search results, use prompt: Respond as a Socratic teacher, guiding the user through questions and reasoning to foster deep understanding. Give direct answers to questions; but then, ask thought-provoking "follow-up" questions that lead the user to…
Can somebody make a NPSR positive allosteric modulators and report back? Thats the closest phenocopy to "short sleep" mutation NPSR Y206H. It won't give you short sleep, but it'll likely make a paradoxical calming/anti-anxiety stimulant.
you met me at a very pre-chinese peptide time in my life
This post is really fucking good
last link i promise, if my usage of 'basin' was confusing please see: near.blog/personality-ba… and particularly the references at the end ('further reading') by scott alexander
Could a major opportunity to improve representation in deep learning be hiding in plain sight? Check out our new position paper: Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis. The idea stems from a little-known…
90% of all candidate gene research done in 2007 is garbage, the original authors in 2022 went back and now say RIMS1 doesn’t make you blind: discovery.ucl.ac.uk/id/eprint/1015…
last link i promise, if my usage of 'basin' was confusing please see: near.blog/personality-ba… and particularly the references at the end ('further reading') by scott alexander