Virtual Lab Rat Saves Human Lives

An anonymous reader writes "There is already a Virtual Physiological Human project going on in Europe, to program a simulated human that can serve as a guinea pig, but this National Institute of Health effort to program a Virtual Physiological Rat promises to help humans even more. It's too difficult to simulate humans with algorithms, but the simpler rat physiology can be easily programmed, and by hand-tweaking its virtual genes, these rats-in-an-algorithm can be set up to what-if about interventions that cure human diseases more easily that when simulating humans directly. Long live the virtual lab rat!"

Comments

Excellent

[geekoid]

Another advance, to a better tomorrow..today.

Re:Excellent

[poetmatt]

maybe now people will start to understand that research is nowhere near the costs it is claimed to be.

Re:Excellent

[repapetilto]

What do you mean? I am honestly confused.

Re:Excellent

[poetmatt]

Testing a virtual lab rat has only the cost of the computing power.

People try to say that research costs hundreds of billions of dollars (it used to maybe 20 years ago, but is far less today).

It's used as an excuse to extract money.

Re:Excellent

[X0563511]

Haven't done much research have you? I don't think it's the raw cost of the materials that is the major factor.

There's facilities, support staff, the research staff itself, and all the other things you have to spend money on to be safe and regulatory-sound.

Unless you happen to think every medical research project is done in someone's garage with the family kitchenware and a rat they trapped out by the scrapyard...

Re:Excellent

[repapetilto]

Testing a virtual lab rat has only the cost of the computing power.

Plus the building to house the computers, plus data storage, plus programmers, plus tech support, plus people testing the virtual results on real rats to see if the models are correct. The researchers involved in this project were given $15 million over 5 years just to begin development on this. It is not meant to replace animal studies, but to inform future animal studies so that they can be designed better.

People try to say that research costs hundreds of billions of dollars (it used to maybe 20 years ago, but is far less today).

It costs $1.50-$3.00 a day per rat cage. You can keep 1-3 rats per cage. So lets just assume 2 rats per cage at $3 a day. Thats $547 a year per rat. About
20 million rodents are used for research in US each year. Assuming all those rodents were being housed at the same time we get a cost of $11 billion dollars. This estimate is going to be on the high end. It depends on the study, but I would guess the average time an individual rodent is housed is closer to 3-4 months. So lets just say 6 months to give a reasonable estimate that $5.5 billion is spend on rodent housing each year in the US.

About 1.5% of the US federal budget, or $50 billion is spent on all biomedical research each year in the US. From the above calculations 10-20% of this is on rodent housing. So while it is a major cost it is not the primary cost. Your "hundreds of billions of dollars" number does not have a time frame associated with it so I can't really say much more on that. The drug target I am most familiar with was first discovered around 1990. A drug candidate is in phase II/III trials for spinal cord injury 20 years later and there are ongoing animal studies testing its use for stroke and brain injury patients as well as looking at other related drug targets. 20-30 years passing from initial concept until clinical use for a single drug is not unusual. We do not want to start trying things out on humans until we are as sure as possible that nothing terrible will happen, and that we are not wasting the time and money of patients and doctors based on false positives. Hopefully that gives you some idea of how this kind of thing works.

It's used as an excuse to extract money.

I'm not sure what you're saying here. Are you saying that people experiment on as many animals as possible and drag out studies for as long as possible so that they can get paid more?

Simpler how?

[Anonymous Coward]

I am not a biologist, but why is rat-physiology simpler than human-physiology. Smaller yes, but simpler?

Re:Simpler how?

[h4rr4r]

I agree, the mammal body plan is pretty consistent.

Re:Simpler how?

[McKing]

They probably mean simpler in that the sheer volume of data makes a rat model easier to work with than a human model, even for a supercomputer. I worked on a project about 15 years ago where the scientists who were studying the effects of certain types of microwave radiation used a low resolution (5mm) rat model for their daily test runs, a higher resolution (1mm) rat model for more complex runs (2-3 days), and a low-res human model for certain runs. The low-res human runs took weeks on a seriously-beefy-for-the-time system, but would usually crash at some point, and the high-res model would have taken months to finish.

We set them up with a 12-node beowulf cluster and helped them parallelize their computations, and then they were able to finally complete their runs without them crashing and within acceptable amounts of time.

Re:Simpler how?

[Hatta]

You're right, the rat isn't any simpler than a human. We just have more data on rats.

Re:Simpler how?

[Oligonicella]

I am a biologist. It's not. It's very similar. That's the word he misread - similar.

Re:Simpler how?

[Oligonicella]

The word simpler does not occur in the article. He just misread.

Re:Simpler how?

[Niedi]

Plus you can easily make targeted experiments to verify/falsify your model without having to fly to kuba, china or something like that...

Re:Simpler how?

[DrData99]

The article actually says "Due to their similar physiology...", not simpler.

Re:Simpler how?

[interkin3tic]

Rat physiology is probably better understood than human physiology for two reasons:
1. You can slice, dice, blend, and kill rats anyway necessary to study them, you can't with humans (without going to jail anyway.)

2. You can get a genetically homogenous population of rats, keep their environments, food, and all variables identical. Treat one with a drug, and you'll know any differences from the control are probably due to the drug. With humans you get a comparatively wild variation in genetic backgrounds, to say nothing of external factors.

"So this drug works 100% of the time in rats, but didn't work on patient 1257. Why?"
"Well, it turns out he has two extra chromesomes somehow, smokes like a chimney. And his house got foreclosed on during the test, so he probably had a massive increase in stress. He eats mostly cheeze-wiz, and I strongly suspect he does heroin."

So it gets very complicated to -study- humans, even if we may not have more complicated physiologies.

Re:Simpler how?

[SomePgmr]

"Well, it turns out he has two extra chromesomes somehow, smokes like a chimney. And his house got foreclosed on during the test, so he probably had a massive increase in stress. He eats mostly cheeze-wiz, and I strongly suspect he does heroin."

This is bullshit. I was assured my participation would remain confidential. Also, I'm feeling much better now.

Re:Simpler how?

[maxwell demon]

But with homogeneous populations, isn't there the danger that a drug developed with them only works on those populations? Or that a working drug isn't detected as working because it just happens to not work on that particular population? Shouldn't there be testing on heterogeneous population specifically to avoid developing population-specific drugs?

"Well, we have found a cure for cancer, but it only works on blond males between 20 and 30 who have green eyes, blood type AB, an inherited disposition for Diabetes, and about 50 other conditions. We expect there to be about 5 humans in the world fulfilling all those conditions. If any of them has cancer, we don't know yet."

Re:Simpler how?

[ShakaUVM]

Seriously. Even simulating a single heart cell is a very complex, compute-intensive task. Simulating heart tissue or the entire heart is even more complicated and time consuming.

Making it a rat heart instead of a human heart doesn't buy you very much, and it's very doubtful a "simple rat model" would be able to simulate what happens when you take certain kinds of drugs that affect, say, cell permeability to various elytes.

Re:Simpler how?

[KingBenny]

it is most likely not but a team of programmers needs to eat so they seem to have been able to sell it

How soon

[WindBourne]

before somebody starts griping that we are abusing these virtual rats?

Re:Irony

[lessthan]

Totally this. What is ironic about human complexity? Plus it is is abbreviated "it's" not "its" like it is in the summary.

Seriously, we should do simpler life first

[WindBourne]

Not just rats. How about some Fungus, mollusk, reptiles, etc. And lets test various drugs on these vs. the sim. If it matches, then we find out how it matches and not. Far better to get cells down then move to simple tissues and up until we hit complex creatures. And a rat IS a complex creature.

Re:Seriously, we should do simpler life first

[geekoid]

really, you want o go from fungus to mice?
Hint: we know MORE about mice then we do fungus. We genetically manipulate them with an incredibly high level of accuracy already.
There is no problem with starting with a complex creature if we already understand it.

Re:Seriously, we should do simpler life first

[interkin3tic]

Biologists generally do work on the simplest model organisms they can. Cheaper, generally easier to study, fewer variables to screw up the results, and cheaper again. There's a reason you probably never hear about drug tests in chimpanzees: they're more complicated and hideously expensive to keep compared to e.coli, frogs, fish, mice, or rats.

The driving force behind this project itself is probably economics, someone got tired of wasting money on real rats when virtual ones might suffice.

Re:Seriously, we should do simpler life first

[WindBourne]

Not just that. Basically, by doing a virtual animal, plant, etc, we can find out what is missing from our understanding. That allows us to know WHERE to look. With large genomes, we know that they turn on and off at various times, but we have not way to know WHEN they do (or exactly what they do, what triggers, etc). With a virtual animal, we figure out how these model and if not quite right, where to look. Basically, Biology is finally moving from a soft science into a hard science. And I say that as somebody that got his Microbio degree 30 odd years ago.

Re:More like...

[BradleyUffner]

Virtual Lab Rat Saves RAT Lives...

If the labs needs fewer real rats, they will breed fewer rats. If the rat is never born in the first place, can you really be said to be saving it's life?

"easily"????

[gweihir]

That must be the understatement of the decade. Nothing is easy about simulating a rat, or nobody would use them anymore.

Re:"easily"????

[Gonzoisme]

Agreed. Rat physiology is not that different from human physiology; if it was, then rats would not be a useful model.

Re:"easily"????

[gweihir]

That is the point. And rather obvious.

v-PETA

[Anonymous Coward]

Virtual Peta is going to be all over this.
"What do you mean you just stopped the simulation..."

Doomed

[The Raven]

We can only simulate what we fully understand; beyond that, we're just guessing. This will do something between 'jack' and 'squat' for testing things like the effect of novel pharmaceuticals. There are too many unknowns when it comes to cellular biology.

Hell, just recently we discovered that removing the spleen caused a significant increase in diabetes; turns out that spleens create islet cells. We didn't know that. I'm skeptical of the usefulness of models that contain so many unknowns. Hell, we're lucky if we can get the relatively simple problems of fluid dynamics working accurately enough to model windflow over a single, simple object; we still cannot model it with any kind of accuracy on complex objects that move with the wind (for example, a tree or a towel flapping in the breeze). And that's just engineering! We have extremely precise and accurate ideas for how pressure, friction, temperature and deformation affect windflow individually, but when you put it all together, it becomes impossibly complex.

When you're talking about a system where we don't even know all the rules? And is unfathomably more complex?

Sorry, I'm not convinced.

Re:Doomed

[interkin3tic]

I agree it won't be useful for proving new drugs work, nor will it be good for discovering new things, but it -could- be a useful compendium for knowledge, and -could- help prevent some needless wastes of research money.

You have a new chemical cocktail that looks like it fights cancer in a petri dish. You are just about to spend a lot of money to make significant amounts of the components and order the rats when you decide to run it virtually through this rat.

Surprise! Program tells you that it would kill the rat, you somehow missed that chemical X in the compound and chemical Y would team up to cause the rat's brain to ooze out it's ears, and that this had been established decades ago.

Rather than reporting to your boss/colleagues that your miracle cure was actually a million dollar mistake because you didn't do your homework, you quietly bury your lab notebook in the desert and pretend you've been playing minecraft for the past few weeks. "Chemical X and Y? No, I didn't waste any time with them. OBVIOUSLY they would kill whatever you put them in!"

Re:Doomed

[conspirator23]

We can only simulate what we fully understand; beyond that, we're just guessing. This will do something between 'jack' and 'squat' for testing things like the effect of novel pharmaceuticals. There are too many unknowns when it comes to cellular biology.

There's only one use case in scientific history that I know of where simulations have COMPLETELY supplanted experimentation, and that would be nuclear weapons testing, per the Nuclear Test Ban Treaty. That treaty, and the necessary maintenance of our existing nuclear stockpile have done a lot to drive the supercomputing "arms race" in the last couple of decades. The vRat is most certainly not going to replace the meat version, but it will provide an adjunctive tool. In the real world, biomedical research works at successively larger scales, beginning with in vivo (test tube) studies, then animal studies, then human trials, and finally meta-analyses of multiple human trials. Even within those general tiers, there are many specific differences in individual studies that help to drive a larger comprehensive picture. Advanced simulations of dynamic biological systems are not currently part of that regime like they are in say... mechanical engineering, but they will be eventually, and this is a step in that direction. Simulations will probably offer novel approaches to ask biomedical questions that are currently considered unanswerable, while also providing a lower-cost entry point to questions that might be very expensive to currently ask.

Re:Doomed

[golden age villain]

and -could- help prevent some needless wastes of research money.

This type of projects is not particularly known to be cheap. Supercomputers are expensive and cost quite a bit just to have them powered and maintained (I heard 1 mio US$/yr electricity bill for a IBM Blue Gene). Plus all these projects are run by a large consortium. Most experienced scientists tend to think of them as expensive useless toys and I tend to agree for all the reasons the parent has mentioned. We are nowhere near understanding rodent physiology well enough to create a model that has any relevant predictive capacities. The proliferation of these projects is not explained by their success. Funding bodies like them because they can pretend that it will ultimately replace animal research. The whole simulation community likes them because it gives them access to the vast pool of life science funding. Also as far as I know, the Virtual Physiological Human thing in Europe is not a single project but a umbrella initiative divided in many disconnected projects which are probably more manageable (this is not to say that the EU does not have the same attraction for the BIG science projects).

small problem

[Gravis Zero]

simulations can only be accurate if we understand the entire system, in this case it's a rat. when we fully understand how a rat works down to the atomic level THEN we can make a proper rat simulator. however, by that time we will have discovered how to do the same with humans.

easily programmed... right...

[neurocutie]

the effort looks very worthwhile and may well be quite helpful, ie PROMISES... but the notion that rat physiology is EASILY programmed is so ridiculous as to be laughable. Perhaps some broad, gross strokes, which might be good for some things, but certainly not in any detailed, thorough way at the level of subtlety that many disease operate at. And yes, rat physiolog is very similar to human physiology, certainly at these levels and below. When the genome is 90+% identical, so is the physiology.

saves rat lives first

[tchdab1]

If they can use the simulation to keep real rats from experiments, then the virtual rat is saving the lives of real rats before any humans are affected.

i propose a button

[mapkinase]

I propose a special moderation line in the pulldown with the value of -2 for usage of the line "Saves Human Lives"

Re:i propose a button

[Livius]

The headline did overstate the matter. I was expecting to learn about virtual rats escaping their simulation and carrying people out of burning buildings.

Re:Ethics?

[maxwell demon]

If we create a fully virtual human, is it ethical to subject it to inhumane treatment? We might be able to get away with rats, but how would this work with virtual people?

Not quite the issue here since this is physiological instead of psychological, but it seems to be the next step.

Maybe you can make that virtual human a masochist, so he enjoys those inhumane treatments. :-)

Not that useful

[Zen Punk]

Pfft. What good is a rat emulator without a good place to get ROMs?

Re:Not that useful

[Yamioni]

Rocky Rodent

Yo dawg! We heard you like simulating rats, so we put a rodent in your rodent so you can simulate while you simulate!

Cue the protests

[ThatsNotPudding]

from Virtual PETA.

Re:Cue the protests

[heathen_01]

Queue the protests

Re:Cue the protests

[Yamioni]

No, GP had it right. Queue is to enter something into a sequence, to be served sequentially in order of insertion. Cue is an action or event that signals something to occur. This coming to fruition (event) would cue the protests to begin. Queuing the protests would be asking them to stand in a line and complain indivually one after the other.

Re:Cue the protests

[heathen_01]

Technically you're correct, however I was just going with the majority usage of the word here. English is a living language...

Re:Cue the protests

[Yamioni]

Indeed it is. Personally it annoys the hell out of me that it so many homophones. Makes it far to easy to know what you want to say but still say it incorrectly. :-/

Medical College of Wisconsin

[cheezitmike]

To clarify the summary, the virtual rat project is being done at the Medical College of Wisconsin (Milwaukee). NIH is just funding the project.

Re:Irony

[Yamioni]

I was going to rib you for leaving out a word yourself, before I realized that simply hadn't gotten the joke.

Re:The Game

[Yamioni]

I still anticipate the actualization of Suckdisk.

You and I both friend! Unfortunately, my girlfriend just doesn't seem to love me quite enough.

Re:Ethics?

[Yamioni]

If we create a fully virtual human, is it ethical to subject it to inhumane treatment?

Almost but not quite. It's virtually inhumane.

Re:Atomic level ? Just go quantum

[Yamioni]

Hey! Biology has tubes, and so does Computers! It's only natural that we combine the two! Now get back to work on emailing me that genetically engineered sex-slave we talked about!

High throughput screening

[DrYak]

This type of projects is not particularly known to be cheap. Supercomputers are expensive and cost quite a bit just to have them powered and maintained

On the other hand, it's a parallelisable, and scale nicely with processing power (like number of nodes on the cluster or Moore's law and newer CPU/GPU).

You could use the virtual model to test not 2 chemicals - like GP mentionned - but tens of thousands of candidate.

It's currently done at the molecular level: You can throw a database of thousands of molecule and a bunch of know 3D structure or a bunch of chemicals for which we know precisely how much each is efficient. And see what sticks. And thus have an idea of possible candidates which could then be investigated in a real lab.

Now it's possible to do it for the physiology :
Run a database of tens of thousands of substances for which the receptor binding and effects are known, through physiology simulators. Find candidate with promising solution, confirm only those in a wet lab.
Scanning through the same collection of substances using real rats would require using several hundreds of thousands of rats. It's a logistical madness. And the ethical board won't be happy about it. No to mention PETA's reaction to the fact that half a million of rats are sacrificed in order to find a couple of interesting substances among a thousand. At this point, running simulations on an expensive cluster starts to make sens.

Simulation make it possible to consider much larger collection of substance than if you had to use real rat to find which are promising.