Supercomputers Assist In Search For New, Better Cancer Drugs

aarondubrow writes: Finding new drugs that can more effectively kill cancer cells or disrupt the growth of tumors is one way to improve survival rates for ailing patients. Researchers are using supercomputers at the Texas Advanced Computing Center to find new chemotherapy drugs and to test known compounds to determine if they can fight different types of cancer. Recent efforts have yielded promising drug candidates, potential plant-derived compounds and new target sites that can lead to more effective drugs. From the Texas Advanced Computing Center: "Identifying a new drug by intuition or trial and error is expensive and time consuming. Virtual screening, on the other hand, uses computer simulations to explore how a large number of small molecule compounds 'dock,' or bind, to a target to determine if they may be candidates for future drugs. [...] In September 2016, writing in the journal Oncogene, Rommie Amaro, professor of Chemistry and Biochemistry at the University of California, reported results from the largest atomic-level simulation of the tumor suppression protein [p53] to date -- comprising more than 1.5 million atoms. The simulations helped to identify new 'pockets' -- binding sites on the surface of the protein -- where it may be possible to insert a small molecule that could reactivate p53. They revealed a level of complexity that is very difficult, if not impossible, to experimentally test. According to Amaro, computing provides a better understanding of cancer mechanisms and ways to develop possible novel therapeutic avenues."

Simulation is too slow

[WrongMonkey]

I used to do this type of research when I was in grad school a decade ago. The problem is that every simulation is really just an approximation. When you take the approximation errors and compound them over a million atoms, you end up with a result that is qualitative at best. OTOH, high throughput screening technologies have been improved to the point where you actually can do trial-and-error experiment on hundreds of thousands of compounds and often get faster than it takes to run the equivalent number of simulation. Methods like cellular thermal shift assays can measure thousands of combinations of drug/protein interactions in a experiment that takes about four hours.

Re:

[SNRatio]

What has happened to Slashdot? I saw the post title and expected to see a big dose of the "Andy Groves Fallacy" in the comments section. Instead I found multiple well reasoned critiques of in silico drug discovery. This place has changed.

Re:

[WrongMonkey]

That's the most cutting insult I've received on /.

Nothing new here

[nbauman]

I remember going to the American Chemical Society conference in 2003 and seeing beautiful molecular models that Merck was using to identify pockets in proteins associated with prostate cancer. They were beautiful. However, Merck never got any useful drugs out of it.

There was a lot of research like that in the 1990s, the most successful of which was the discovery of imatinib (Gleevec) which essentially turned chronic myelogenous leukemia from a fatal disease with a 3-year median survival into a chronic disease with a >95% survival. Of course they studied the binding pockets with computer simulations. It's been a long time since they cut electron density maps out of plastic and stacked them up. But they used to do it without computers, and they could do it again.

As best as I can figure out, the Texas Advanced Computing Center put a shitload of money into a new big fucking supercomputer. They can do the same thing, only faster. Sure, they might come up with a cancer drug.("Put it in the grant that we can use it for cancer research.) But in the best of times, they have some pretty high odds to go through before they get to anything useful. It's a good thing they have a high throughput, because with those odds they need it. The interesting thing is that they actually found a drug, mebendazole, that's already used for parasitic worms that might, maybe, no promises be a candidate forfurther exploration in the treatment of cancer. There are easier ways to find it, but fine. (One of the problems with anti-parasite drugs is that the metabolism of parasites is so close to that of humans that it's hard to find drugs that fuck up worms without also fucking up the host (us).

But great, keep trying. In general, it's great to spend money on basic research tools. How's that cancer moonshot going?

Now I have to admit that I have a weak spot for spending shitloads of money on supercomputers, for modeling cellular molecules and their interactions. Some people get hard from shooting up x-ray telescopes, and if that gets you off, I hope you get a billion dollars (or Renminbi, if that's where you get your grants) for it. Personally, I can watch p53/DNA pornography all night.

But as far as I can figure out, this is just routine incremental research. Not that I mind. In fact, I'm glad for the TACC PR department that wrote this press release, because now they can show their bosses that they got pickup in Slashdot, America's premier news for nerds site -- hey, wait a minute. The guy who submitted the story, Aaron Dubrow, is the same guy who wrote the press release for TACC. Oh well. Well played.

Anyway, I'm off to find cell biology videos. https://www.youtube.com/watch?... [youtube.com] And then I want to check out Dance Your PhD.

Re:

[methano]

I've been watching and dabbling with this modeling stuff since the early 80's. I tend to agree with you. This is big spending along a rather shallow incline. Not so good that I'm gonna start smoking again.

About Gleevec, I don't think modeling had a lot to do with it. Identifying the target, and running a big screen were more important in its discovery.

Results of the @home 'supercomputer' projects

[Modern]

There are the @home projects. Are these not supercomputer-like projects. How much fruit falls from those trees? Is a single supercomputer going to do much better than that?

Soon, what will kill more people than cancer?

[modmans2ndcoming]

Bacterial infection will kill more people per year than cancer soon because of our poor stewardship of antibiotics. We should be putting way more effort into solutions for infections (production as well as research) than we do today.

Re:

[loranger]

Bacterial infection will kill more people per year than cancer soon because of our poor stewardship of antibiotics. We should be putting way more effort into solutions for infections (production as well as research) than we do today.

A lot more people get killed by car than by swimming pools. It's time to stop this mandatory swimming-pool-protection-fences nonsense, and focus on car safety instead!

Have you ever hear about the law of diminishing returns? Read it, and think again.

Re:

[modmans2ndcoming]

1) you assume we put any real production research into new antibiotics
2) Given how poorly people manage antibiotics around the world we need more basic research funding just to keep up.

funny..

[SuperDre]

supercomputers have been used for years, trillions of dollars have been poured into cancer research.. and we all know there already are real cures to cancer, but it's still more profitable to design/sell drugs for reducing cancer instead of curing it. So the actual cures are locked in some vaults and only available to a very small portion of people. what's the use for a pharmaceutical company to actually produce a cure, which would cure everybody and that's it, there it ends, no more extra income..

Re:

[the gnat]

Every time an article about cancer therapies (or pharmaceuticals in general) comes up, someone always makes this claim. Aside from the sheer misanthropy of assuming the worst of everyone involved in drug development, it's spectacularly ignorant of human biology, modern medicine, and the pharmaceutical business. The truth is that most drug candidates end up being colossal wastes of money with nothing to show for it, usually because of poor efficacy in large-scale clinical trials (but occasionally due to un