16 ms·
I assume you mean computer models? The problem here is that, although it might be possible to figure out a drug's efficacy by modeling just it's primary mode o
by GlenTheMachine 8y ago
I assume you mean computer models?
The problem here is that, although it might be possible to figure out a drug's efficacy by modeling just it's primary mode of action, in order to find it's side effects you have to model every other interaction it might have.
In other words - we can tell whether a drug will kill cancer cells just by looking at how it interacts with cancer cells. But to figure out how it interacts with the rest of the body, you have to be able to model the entire body with nearly perfect fidelity. And this is not just a problem of having insufficient knowledge - although it is that - we just don't understand all of body's systems anywhere near well enough to write such a model) but it's also a compduting power - recall that the protein folding problem is NP-Hard with an enormous number of degrees of freedom. In some cases it can have several hundred degrees of freedom, and the number of cases you have to check scales as the factorial of that.
Maybe someday we'll have numerical techniques that can solve the protein folding problem and other organic chemistry problems using achievable computing power. But we can't today. Even if we had a perfect atom-for-atom model of the human body we couldn't run it.
- ethbro 8y agoOr, to put it another way, why would drug companies be spending billions of dollars on animal model trials if they could simulate drugs in silicon? A teaspoon of humility when talking about why another discipline doesn't 'Just do X?' is well served.
- ianlevesque 8y agoI read it more as an “if I had a magic wand” than judgement. How cool would it be?
- choeger 8y agoYou might have misread my comment. I find it sad that we do not have these models. I am well aware that right now it is not possible. I do blame on pharma however for not putting massive amounts of money into this fundamental research. I did a lot of modeling and simulation and I know that chemistry and molecular biology simulation tools exist. But I never encountered a call for academic contributions to increase performance or supply general numerical or computational techniques. from my pov the civil engineering simulators (using odes,dates, and pdes) where completely separate from the chemistry world (often using odes and daes).
- GlenTheMachine 8y agoI can tell you that there is a not inconsiderable amount of effort going into the computational protein folding problem. I do not know where this funding comes from though. It's just a fundamentally hard problem, and it isn't clear (to me, at least) that it can really be solved. Quantum computing, maybe, but I'm not knowledgeable in that. Short of that, though, it might always turn out to be computationally more efficient to just "build" an analog body. There's no guarantee that we can simulate every physical process at faster than realtime speeds with CPUs. I don't think you're going to find a single line of accounting that says "improve biochemical knowledge of the body for modeling purposes". Essentially, all fundamental research in medicine falls into this bucket.