new blogpost about using a CGM for 30 days and which foods did what to my blood glucose blog.booleanbiotech.com/blood-glucose-
Pinned Tweet
Follow
Brian Naughton
@btnaughton
Joined January 2015
Brian Naughton’s Tweets
Great new article on technologies for drug target ID from featuring labeling approaches, MS-based label-free approaches, etc drughunter.com/resource/metho
14
Great data science video series, with PCA, importance sampling, gaussian processes, etc. e.g. youtube.com/watch?v=lJ0cXP
1
15
Computational approaches streamlining drug discovery
25
119
Here's the main take-home message:
I made a tool that combines instantaneous ESM structure predictions with Web-based molecular graphics to rapidly explore the potential impact of mutations.
Try it here!
mutationbrowser.com
1
4
13
Show this thread
Today we report that an engineered skin bacterium, swabbed gently on the head of a mouse, can unleash a potent immune response against a distant tumor. led the charge w/ help from , , . 1/26
40
521
1,493
Show this thread
Worthwhile book review / history of IRBs on astral codex ten astralcodexten.substack.com/p/book-review-
2
interesting conclusion from the EigenFold paper, which generates a distribution of structures from sequence arxiv.org/pdf/2304.02198
1
2
12
Is anyone using this in production? It would be great if it was just as good as PCR, but isothermal.
Show this thread
Interesting blind docking results from arxiv.org/pdf/2303.09916
Vina / SMINA / GNINA seem to perform well here compared to DiffDock.
2
7
48
Is there any project where they try to humanize as many mouse (or other model organism) genes as possible? For mouse, I can mainly find immune system replacement.
1
This is the first time I've seen self-driving that seems close to working -- pretty sweet!
1
is this how kids will learn to program?? from zero to a basic javascript game in a few minutes
1
4
1/24 It is very interesting to watch come to the rescue of the fake certitudes of a market zoonosis, in the form of the highly controversial 'dispositive' evidence pushed last year in the New York Times and then in Science.
He was in overdrive over 24 hours:
12
73
166
Show this thread
New blogpost about using 's Paper QA / GPT-3 to create a knowledge-base of scientific papers for a lab or biotech.
blog.booleanbiotech.com/gpt3-as-biotec
3
10
57
I used pdfminer and GPT Index (now LlamaIndex) to index a set of ~200 papers. When I query, the results are much worse than chat-gpt, including not knowing nouns directly discussed in the papers and present in the index json many times. Anyone know what I am doing wrong?
1
2
We are excited to announce our Series B raise!
3
44
the answer to the question of how salamanders breathe is pretty interesting! turns out the can use lungs, gills, absorb oxygen cutaneously, or various combinations of these? animals.sandiegozoo.org/animals/salama
4
1
2
16
Sounds like you need to buy the (upgrade) stainless steel attachments to avoid this issue.
2
Show this thread
"Lead safe mama" has some interesting article about lead testing. I hadn't heard of X-Ray Fluorescence... tamararubin.com/2016/12/ask-ta The KitchenAid Mixer stuff is very concerning: tamararubin.com/2023/01/urban-
2
1
4
Show this thread
Replying to
“How to Survive a Plague” by David France, 2016
1
1
2
A while ago I read an excerpt where an experimental treatment was tested on a scientist, and the scientist was listed as a "primate". It thought it was a HIV treatment, maybe at Merck... Does anyone know the reference?
1
1
jupylite experiment: make 's github.com/fogleman/sdf library run in the browser (remove multiprocessing, make wheel...) s3.amazonaws.com/booleanbiotech
The jupylite experience is still a bit buggy, but it's pretty cool that pip works!
1
Upcoming home urinalysis "smart toilet" from withings: withings.com/us/en/u-scan Based on the patent patentimages.storage.googleapis.com/f1/d6/f1/26824 it uses something like ISFET/EnFET/Bio-FET en.wikipedia.org/wiki/Bio-FET and probably tests 3-4 metabolites at a time.
1
1
Introducing the Spinout Playbook, a free guide to spinning a startup out of academia.
fiftyyears.com/spinout
Why is this needed? The process of “spinning out” science from universities is broken 👇🧵
GIF
14
241
650
Show this thread
related
1
Show this thread
Meet PubMed GPT 🩺 a new SOTA on the US Medical Licensing Exam developed by MosaicML and . It's a normal GPT-3B model trained on medical data that bests hand-designed med models and generic models 40x bigger, a sweet spot for foundation models🧵
11
157
532
Show this thread
2022 H1B data in pandas (slow dl!)
df_h1b = pd.concat([pd.read_excel(f"dol.gov/sites/dolgov/f{q}.xlsx") for q in [1,2,3,4]])
(df_h1b.loc[lambda df: (df.VISA_CLASS == "H-1B") & (df.FULL_TIME_POSITION == "Y") & (df.WAGE_UNIT_OF_PAY == "Year")])
2
All the gpt3 chat stuff reminded me of this paper where they compare prompts to try to find the one that makes gpt3 smartest. So weird…. arxiv.org/pdf/2205.11916
1
1
2
I have no idea what is going on here, but it's interesting. There is no CHANGES.md in the current minimap2 repo, but the changes listed sound very plausible. They also extend into 2022, past the date the model was trained.
1
3
Did you know, that you can build a virtual machine inside ChatGPT? And that you can use this machine to create files, program and even browse the internet?
245
3,010
8,520
Show this thread
Approval for molnupiravir (MOV) use should be revoked worldwide. Not only has it proven ineffective, it seems clearly to be creating sequences such as the one below, which in ~2 months acquired 72 mutations, more than its lineage had accumulated in ~3 years of evolution. 1/
77
766
1,847
Show this thread