Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Supercomputer analysis of Covid-19 leads to new theory

elemental.medium.com

Post reply on HN
Print view View original

111–120 of 154 posts

Re: Supercomputer analysis of Covid-19 leads to new theory

#111

Post by interestica »

afarrell wrote: post #101
klysm wrote: post #3

The anthropomorphization of computers (especially powerful ones) is pretty annoying. A _person used a supercomputer_ to analyze Covid-19, and I’m guessing the theory didn’t just ‘emerge’ from the computer.

People like conciseness. "A person used a supercomputer to analyse Covid-19" conveys no more knowledge than "A supercomputer analyzed Covid-19".

Also, it prevents one from having to get into too much detail: was it a group of persons? were other tools used? What does 'analyse' mean in this context?

Re: Supercomputer analysis of Covid-19 leads to new theory

#112

Post by secabeen »

dekhn wrote: post #13
secondcoming wrote: post #7

Earlier quoted context omitted.

Maybe to test the algorithms and results, so that next time something like this happens there can be a list of suggested treatments ready far sooner.

The issue is that, from what I can tell, the authors just used R to analyze some data, with no explicit parallelism. You would do a better job just renting time on AWS, saving money for everybody.

In my experience with research computing, if you are able to keep a computer doing active work more than 60% of the time, it will be cheaper to purchase and run that computer yourself than renting it from AWS. That's the case even with commodity machines with only 10G Ethernet interconnect. $15k for a machine is only $0.34/hour over 5 years. That doesn't buy much of an AWS machine. (Yes, cooling, real estate and power are all overhead on that, but researchers often don't pay those costs directly, they are covered by the university with other monies.)

Re: Supercomputer analysis of Covid-19 leads to new theory

#113

Post by woeirua »

michaelhoffman wrote: post #63
woeirua wrote: post #29

Seems like it should be really straightforward to determine if this hypothesis is right by measuring bradykinin levels in Covid patients versus non-Covid patients. One thing that I think strongly suggests that this hypothesis is wrong is that there is no strong relation between ACE-inhibitors and Covid mortality. Indeed, most of the studies that I've seen suggest that ACE-inhibitors have a somewhat protective effect…

The cited article has been retracted. (I know this because I have the PubPeer extension installed which puts a big red warning by it.)

My apologies. I missed it earlier. Here is another article that looks to be in good standing that comes to similar conclusions about ACE-I vs ARBs [2].

[2] https://www.nejm.org/doi/full/10.1056/NEJMoa2008975

Re: Supercomputer analysis of Covid-19 leads to new theory

#114

Post by dekhn »

secabeen wrote: post #112
dekhn wrote: post #13

Earlier quoted context omitted.

The issue is that, from what I can tell, the authors just used R to analyze some data, with no explicit parallelism. You would do a better job just renting time on AWS, saving money for everybody.

In my experience with research computing, if you are able to keep a computer doing active work more than 60% of the time, it will be cheaper to purchase and run that computer yourself than renting it from AWS. That's the case even with commodity machines with only 10G Ethernet interconnect. $15k for a machine is only $0.34/hour over 5 years. That doesn't buy much of an AWS machine. (Yes, cooling, real estate and powe…

You're completely ignoring the other valuable aspects of being in a cloud: you're close to huge amounts of high throughput storage (blob and DB), and can increase/decrease the size of your fleet trivially. These are critical to nearly all modern scientific workflows (moreso than the raw compute, IMHO).

As for the cost structure for research computing, the argument that the costs are externalized isn't a good one- that overhead that pays for the facility, and the networking, comes out of your grant money, and using grad student time to admin your cluster often just causes your grad students to leave for FAAMG.

Re: Supercomputer analysis of Covid-19 leads to new theory

#115

Post by 08-15 »

zby wrote: post #23

They found out that some genes related to ACE in comparison to genes related to ACE2 are more 'expressed' in Covid patients than normally and they conclude that that must have resulted in too much bradykinine. Hmm - that strikes me as kind of roundabout - why they couldn't just measure bradykinine levels directly? Is that too hard? "Here, we perform a new analysis on gene expression data from cells in bronchoalveolar…

> why they couldn't just measure bradykinine levels directly? Is that too hard?

It is. To measure gene expression, you isolate total mRNA and sequence it. This tells you the expression of all genes simultaneously. The protocol is fairly standard, cheap, and quick. That doesn't tell you anything about bradykinine, though, because there is no mRNA that codes for it.

In contrast, no such protocol exists for proteins. Sequencing a single protein is comparatively difficult, and no high throughput device exists that sequences lots of proteins, let alone quantifies their abundance. The traditional lab methods like PAGE gels are slow and labor intensive.

Re: Supercomputer analysis of Covid-19 leads to new theory

#116

Post by searine »

tzs wrote: post #56
searine wrote: post #11

Earlier quoted context omitted.

>I can't really see why you would need a supercomputer to do the analysis they did The article is fluff. 98% of all bioinformatics is done on "supercomputers" or 'high performance computing environments" saying the researchers used supercomputers to analyze the expression data is like saying someone used a shovel to dig a hole.

The article doesn't simply say they used a supercomputer. It says they used Oak Ridge's Summit supercomputer, the second fastest in the world. Unless 98% of all bioinformatics done on a top 2 supercomputer, your point fails.

>It says they used Oak Ridge's Summit supercomputer, the second fastest in the world.

The shovel had racing stripes. Got it.

The job doesn't change based on what hardware it is run on.

Re: Supercomputer analysis of Covid-19 leads to new theory

#117

Post by sydd »

ChrisArchitect wrote: post #60

it's annoying that this is not a 'new' article/theory. It is an article talking about some research and analysis from TWO MONTHS ago. Almost want to request a (July 2020) marking in the title as we do with old link posts

Actually the theory is even older, here's an article from the beginning of April: https://www.radboudumc.nl/en/nieuws/2020/radboudumc-research...

If this info has been out since 5 months I'd expect several trials already that target the bradykinin storm.

Re: Supercomputer analysis of Covid-19 leads to new theory

#118

Post by SideburnsOfDoom »

s1artibartfast wrote: post #73
Jtsummers wrote: post #16

Earlier quoted context omitted.

> theories are still a purely human domain. Why do you believe that? Theories are attempts to explain the mechanism of something based on the observed data. Given data on patient symptoms and known drugs (ACE inhibitors in this case) and their effects, a computer could easily produce a theory that the disease acted like ACE inhibitors. It'd still take a human to write the program to generate these theories, but a com…

I think it is a question of semantics regarding the words generate and theory. Computers cant produce theories because they generate and analyze data, but cant process ideas. Unless the computer is conscious, it can not generate the theory.

> they generate and analyze data, but cant process ideas

But what if the data represents ideas?

https://en.wikipedia.org/wiki/Automated_theorem_proving

> Unless the computer is conscious, it can not generate the theory.

Why would you say that consciousness is necessary for theory generation? It isn't for arithmetic, equation solving, natural language processing or image identification, etc.

Re: Supercomputer analysis of Covid-19 leads to new theory

#119

Post by SideburnsOfDoom »

glaberficken wrote: post #35
SideburnsOfDoom wrote: post #25

Earlier quoted context omitted.

A set of correlations identified automatically in the input data - is that a theory, or merely output data?

Exactly what I thought. A theory has to make a falsifiable prediction. A correlation is not a falsifiable prediction.

> A theory has to make a falsifiable prediction. A correlation is not a falsifiable prediction

Yes, you're correct. However, identifying the correlations is a necessary precondition to making a theory about them. And the automated analysis can help with that step.

Re: Supercomputer analysis of Covid-19 leads to new theory

#120

Post by raverbashing »

WhompingWindows wrote: post #89
raverbashing wrote: post #34

Earlier quoted context omitted.

Yes but can it kill viruses and bacteria in vivo?

By "it" what are you referring to? Vitamin D may have modulatory effects on the biochemical pathways linked in the original article. Most pharmaceutical products we study aren't going to directly kill bacteria or viruses, usually they have a particular physical effect on a specific molecule that then leads, through various cellular pathways, to the death or suppression of the pathogen. I highly doubt Vitamin D direct…

I mean UV light/being outdoors

Yes, I'm aware of UV sterilizer, but does it work when the microorganisms are inside you? You don't want to get an "internal tan"

Post reply on HN