nik_s

The mitochondria in many cancer cell behave abberantly. For example, in many cases, they can start promoting glycolysis, in process known as the Warburg effect. They're also involved in other processes such as apoptosis and cell migration. I'd hypothesize that these drugs target pathways or processes that are dysregulated in the mitochondria of those cancers.

I think this is a very cynical take. Patients with gastric cancer are often in a dire situation: many of them have had their stomach removed and have generally significant difficulties with digestion, which makes consuming specific foods very difficult. The therapeutically effective doses of ATRA are also high compared to what you'd get from foods, meaning that injections are nearly the only solution for these patients. Finally, naturally derived compounds generally have the weakest patents, so if anything, we should thank these researchers for looking into the therapeutic potential of these drugs. I can assure you that few pharmaceutical companies would research this, as the commercial incentives are really weak to do so.

I work in drug discovery (including in oncological indications), so it's always great to see new research with early, promising results!

To summarize the paper: some gastric cancer models are sensitive to all-trans retinoic acid (ATRA). ATRA is an active metabolite of vitamin A, and is already used clinically to treat some hematological cancers. The researchers tested the compound at several concentrations on a number of gastric cancer cell lines, where they show moderate to significant reductions in growth (from 30% to 60% reduction). Next, they tested ATRA on xenografted tumor cells on immuno-compromised mice, where they also show a reduction in tumor size. They hypothesize that ATRA exerts its effects through immuno-modulation.

Firstly, very cool, and congrats to the authors - I definitely see first evidence of the potential of ATRA to treat gastric cancer.

While I might have misunderstood some parts, I do think there are some elements that warrant precautions here: 1. we can see that some of the untreated mice also show reduction in tumor size, albeit less significant, meaning there could be issues with the cells or protocol, 2. I find it hard to conclude anything about ATRA exerting its effect through the immune system in a study of cell models and immuno-compromised mice.

Nevertheless, given the frankly poor prognosis of gastric cancer (many patients in early stages of the disease now get their stomach removed, and more often than not, these tumors metastize aggressively), and the well known safety profile of ATRA, I think the study is very welcome, and should warrant further investigation. Given the hypothesized mechanism of action, I think testing on humanized mice, using more and more distinct patient-derived cells would bring convincing proof to move ATRA forward to clinical trials.

While I agree with the general sentiment of your message, there are quite a few meta-analyses on pubmed on the long-term safety and efficacy of gastric bypass surgery. They mostly show that while some patients do end up back obese, the majority do improve on many markers of health (weight, diabetes, ...).

Here's an example of a study if you're interested [1]!

[1] https://pubmed.ncbi.nlm.nih.gov/29340676/

It can do both - I've edited my original post to show the translation task.

I just tested the model [1] using an RTX3090, trying to translate a french text I found here [2].

Some observations:

- The full translation of the 6:22 minute video takes about 22 seconds (17x real time)

- It recognizes the language by default (and did a good job to recognize it was french audio)

- MIT License [3]!

- The quality of the transcription is good, but not perfect.

- The quality of the translation (if you don't consider transcription errors as a translation error) is generally very good.

---

The transcription:

> Bonjour à tous, <error>j'suis</error> espère que vous allez bien, c''est ENTI. Et aujourd', <error>aujourd',</error> on se retrouve <error>un peu physique</error> pour parler de la termo dynamique. Vous ne vous inquiétez pas, ça va bien se passer. On va y aller ensemble, <error>être à par exemple,</error> je vous accompagne à travers une série de vidéos pour vous expliquer les principes de base en termo dynamique. Et bah, c''est parti, on va y aller tranquillement. Lidée, c''est vous puissiez comprendre la termo dynamique dans son ensemble. Donc, je vais vraiment prendre mon temps pour <error>couplisser</error> bien comprendre les notions,

The translation:

> Hello everyone, I hope you're doing well, it's NT and today we find ourselves a little physical to talk about the thermo dynamic. Don't worry, it's going well, we're going to go together and be the same. I'm going to accompany you through a series of videos to explain the basic principles in thermo dynamic. Well, let's go, <error>we're going to go quietly</error>. The idea is that you can understand the thermo dynamic <error>in sound together</error>. So I'm really going to take my time to understand the notions,

---

All in all very happy that OpenAI is publishing their models. If Stable Diffusion is any guide, people will hack some crazy things with this.

[1] https://github.com/openai/whisper [2] https://www.youtube.com/watch?v=OFLt-KL0K7Y [3] https://github.com/openai/whisper/blob/main/LICENSE

This [1] study was pretty thorough on different fasting lengths. They tested 1422 patients for fasting lengths between 4 and 21 days, with a maximum calorie consumption of 200-250kcal and a moderate-intensity lifestyle program.

It concludes that all fasting lengths are beneficial, and are likely going to result in a) reduction in weight and waist circumference, b) beneficial effects on blood lipids, regulation of sugar and other blood-related parameters, including lower blood sugars and higher ketone body levels, c) an increase in physical and emotional well-being and absence of hunger, d) a high probability of decrease of pre-existing health-complaints, e) very limited chance of side-effects.

[1] https://journals.plos.org/plosone/article?id=10.1371/journal...

You can order these drugs in bulk from chemical vendors for research purposes (i.e. non-human use). Chemscene, to name just one, sells Rapamacyn at 1440 USD for 5g.

EDIT: I clearly don't recommend you buying this and taking it by the way. You're never quite sure about the purity nor about how to dose this effectively as a layperson without access to a lab. It's also possibly completely illegal, depending on your local laws.

Regarding the biology:

- One of the strategies in drug development is to find a protein that is more present in people who have a disease versus those who don't (called "overexpressed" proteins), and attempt to stop this protein from work correctly (called "inhibition") by having a small molecule drugs that binds to a specific site of the protein (called the "allosteric site") in order to make it mechanically unable to execute its function.

- Cyclin-dependent protein kinases (CDKs) are a family of proteins that seem to play important roles in controlling cell division. CDK20 is overexpressed in a number of cancers.

Regarding the novelty:

- Discovering new inhibitors of proteins based on AI is definitely less novel than it was 5 years ago - while it's definitely still not the norm, AI is making big waves in the pharmaceutical industry. However, I think this might be the first publication validating the use of Alphafold for small molecule drug development, which is a major step forward.

- While it's interesting to see that it's possible to design a small-molecule CDK20 inhibitor, it's currently still very uncertain whether this is a promising drug: i. the compound could be insufficiently specific to CDK20 and could bind to other important proteins and cause unwanted and potentially serious side-effects, ii. the compound could have bad "drug-like" properties (e.g. bioaccumulate in the liver) or be toxic in some way, iii. the compound could interact badly with other drugs that cancer patients receive, iv. the compound could induce resistance (a common problem in small molecule drugs in oncology), and finally, and most importantly, v. the drug might just not be effective at treating cancer or any other diseases - it's not because a protein is over-expressed that it's the cause of a cancer, but rather a symptom of another biological dysregulation.

Still, it's definitely an achievement, and I applaud the efforts of the team and hope they'll find successful treatments.

You can also say that blockchains have mechanisms that prevent a government from coercing third parties from handing over your NFTs to someone else.

I'll agree that the analogy wasn't perfect though - I added an edit for another analogy (still as imperfect as it may be).

I'm not sure I agree with your explanation.

Most NFTs are like a title to a house. Like a title to a house, you can: 1. Prove you own it; 2. Sell it to someone.

Unlike the title to a house: 1. The party certifying your ownership is not the government, but the consensus rules of a blockchain; 2. The thing you own is not physical, but digital in nature; 3. Your ownership does not prevent anyone from copying the file itself for their own uses.

Regarding your point that the object you own might not be consistent: because of the cost relative of storing data on a blockchain, the actual digital thing you own isn't usually stored on the blockchain, but a hash of that file (e.g. [1]). Unless hashing is broken, this is cryptographic proof that the object does remain consistent.

[1] https://github.com/larvalabs/cryptopunks/blob/master/contrac...

Edit: perhaps the notion of an entry ticket to a concert would have been a better analogy than a title to a house.

I worked in the same co-working space as Wietse, Aline and Koen, the three co-founders of Citizenlab, when they were launching their service. They're a super competent team with a fantastic vision and execution - I'm excited to see this new step in their story.

A minor point about the podcast: Alex Hern misunderstood the process that Craig Wright, the self-processed founder of bitcoin (Satoshi Nakamoto), needed to go through to prove that he was indeed Satoshi. Alex says that Craig needed to "move the first bitcoin ever created", while in reality Craig needed to sign a message with a private key. Because of how bitcoin is made, the very first bitcoin block (the "genesis block") cannot be spent [1].

[1] https://www.investopedia.com/terms/g/genesis-block.asp

The author implies that the recent growth in tether's market cap is an indication of an incoming exit scam.

A cursory glance at Coinbase's USDC's market cap [1] shows that it too is growing at almost exactly the same pace as tether's [2]. I think most players in this market would agree that Coinbase, for all of it's failings, is unlikely to be planning an exit scam at this point.

It doesn't disprove the whole thesis, and some elements of it might have merit, but at least one element of it seems weak to me.

[1] https://coinmarketcap.com/currencies/usd-coin/

[2] https://coinmarketcap.com/currencies/tether/

Incredibly interesting research: hopefully this pans out in human studies too.

If I understand correctly, in order to achieve this:

  1. a nanotube is created that contains a chemical compound (SHP1i) that inhibits the the "don't eat me" (antiphagocytic) signal of plaque.
  2. These nanotubes are engineered to be taken up by white blood cells.
  3. White blood cells tend to be more present at areas of inflammation (such as plaque), which in turn increases the chance that the chemical compound mentioned above will be released in areas where plaque is present.
  4. This presence of nanotube-loaded white blood cells at areas of plaque will turn off the antiphagocytic signals of all cells in this area,
  5. which induces plaque (and all other antiphagocytic cells in the area) to be removed through traditional cellular activity.

Assuming what I said above is correct, what I still don't understand:

  - How do the nanotubes bind to the white blood cells? Is there a specific technique to achieve this?
  - If I understand correctly, the chemical compound (SHP1i) removes antiphagocytic signals indiscriminately, which causes issues with toxicity of most therapies that attempt to use this technique without tightly controlling delivery (because organs might become a target of white blood cells). What happens however if a patient has other areas of inflammation? Does this exclude patients with auto-immune diseases from benefiting from these approaches, for example?

Any insight would be greatly appreciated!

Other sources: [1] https://www.nature.com/articles/s41565-019-0619-3 (Original article)

I totally understand the sentiment, and generally, when I do solo projects or projects with small teams, I also prefer light frameworks.

In my experience, frameworks like Django and Rails shine when you have to collaborate with large teams on solutions though - the fact that there's comprehensive documentation, best practices, and clear standards make it so much easier to work together on a project. I do think Rust brings a lot of assets through its type systems when collaborating, but rolling your own auth/admin dashboard/orm is a hefty investment for most mid-size companies.

Congrats on the release! Actix is super exciting, and it's great to see how the community stepped up to maintain this amazing project.

Is there any effort to bring a Rails/Django like framework to Rust? AWWY [1] seems to indicate that there isn't anything like it in the Rust ecosystem, but I'd be curious to know if there's anyone working on something in that direction. All of the frameworks I've seen seem pretty "bare-bones" by the standards of Django/Rails (then again - it's hard to beat these frameworks in terms of feature-completeness).

[1] https://www.arewewebyet.org/

On top of that, the author uses aiopg rather than asyncpg[1] for the async database operations, even though asyncpg is (allegedly) a whole lot faster.

1. https://github.com/MagicStack/asyncpg

I'm the CTO at a data science company, and this has been my experience too. I've been lucky enough to have quite a few engineers go from zero practical experience to being able to train and deploy complex ml solutions, and the most successful solutions have always involved a combination of just a couple of tools: - airflow and/or celery for running data extraction and transformation jobs - pandas and numpy for data wrangling - sklearn, xgboost, lightgbm, pytorch or tensorflow for training/inference - flask or Django to serve results

It's a handful of technologies, but they're (generally) mature, battle tested, and well documented.

Very nice! For those who are just skimming the link, this is actually a demonstration of the capabilities of a python package that is available on github [1]. If I understand correctly, most of the transformations are aimed at time-series data.

I'm not sure that saying that augmenting tabular data is novel, as claimed in the packages readme, but some transformations definitely were new to me (e.g. the columnar Gan), and this should in no way diminish the fact that such a library is really very, very handy.

Thank you to the author for the hard work and sharing this code!

[1] https://github.com/firmai/deltapy

Thank you - I came to the same conclusion while still editing my post - I think my edit and your post coincided. I clarified this edit in my original post.

I think you're referring specifically to the "political and cultural pressure in China" part of my original post. I was basing this on reporting from the New York Times [1] of the original outbreak. It very well might be that the reporting has improved since then, but I do think that there is evidence to say that there was manipulation of data, or at least a very strong incentive to under-report cases initially.

[1] https://www.nytimes.com/2020/01/30/podcasts/the-daily/corona...

If my original post gave off the impression of being conspiracy-theory-driven, I want to apologize, because that wasn't the intent.

My original post aimed at clarifying that death rates in exponentially spreading diseases that have a lag between infection and possible recovery and death is a tricky topic, and that the numbers generally reported seem to not take this nuance into account.

I think collecting accurate data in the middle of a crisis is incredibly difficult for a host of reasons, none of which need to originate from ill intent. Test kits may be missing, data may be collected differently, methodologies for evaluating "who is infected" and "who has recovered" may differ between countries, patients' records might just not be a top priority for hospital staff, ...

> Your formula has the opposite problem, by looking at cases that have a known outcome you are biasing results towards cases that end quickly (in particular case that end in death).

Unless I'm mistaken, that would only be true if cases that end in death are much faster than the cases that end in recovery. I have found it difficult to find detailed statistics on the topic, so I cannot tell you that it's definitely true or not, but I agree that it's an important variable that should be taken into the equation.

Indeed - thank you, I adapted the formula in the original post.

Indeed, you are right - I corrected the formulas.

The general issue with this measure of mortality (dead/(infected + dead)) is that you're assuming that the infected won't die. In a disease that is exponentially growing, a better approximation of evaluating your survival chances is to look at the death to recovery rate (dead / (recovered + dead)). Based on the available data [1], we are closer to 7.8% than 2% mortality, which is closer to the final mortality rate of SARS of 9.6% [2].

Nevertheless, I think all these statistics need to be taken with a grain of salt - I doubt that the numbers we are seeing are of very high quality, given the political and cultural pressures in China to underreport, the lack of test kits, corona virus deaths being attributed to other diseases, ... On the flip side, it's very likely that mild cases will never be reported, which in turn would decrease the mortality rate.

[1] https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.h...

[2] https://en.wikipedia.org/wiki/Severe_acute_respiratory_syndr...

[edit: corrected the formula - thank you @anhner and @11thEarlOfMar for spotting the mistake]

[edit: indicated that there's also a chance of under-reporting mild cases - my edit coincided with @Tenoke's post - sorry for noticing this late @Tenoke]

Actually, multiple orders of magnitude...The Flu has a mortality rate of 2 per 100k [1].

[1] https://www.cdc.gov/nchs/fastats/flu.htm

[edited - reformatted numbers to avoid localization issues]