Simulation of life using chemistry?

[KSK]

Wasn't that called Spore.  

Rules for association is going to be a tough one for sure. For a reasonably well ordered protein with a lot of secondary structure so that, to a reasonable approximation, it has a defined shape, it might work. For a disordered protein (I think of it as having a lot of random coil but I suspect my terminology is out of date) it's going to be a bear. My PhD was in self-assembling organic systems - very simple, small molecule systems at that. We worked with hydrogen bonds mainly, because they're relatively strong and directional - and even then there was usually a better than even chance that all the other intermolecular interactions would combine to swamp our carefully designed hydrogen bond network. So yeah, association rules for proteins sounds like a challenge all by itself to me. Although I defended my PhD well over a decade ago, so my knowledge is correspondingly out of date.  

I think we can all agree that an atomic scale model isn't feasible, so I agree with @todofwar that you'd need to find a level of abstraction that works and run with it. The problem with that is, that we're not going to be learning anything new about protein-protein interactions or the fine details of what goes on in a cell. The model is only ever going to be as good as the information we can put into it. If we don't know that kinase X has a subtle effect on isoform Y of protein Z - but only once Y has already been phosphorylated by kinase W, then we're not going to capture that interaction in our model. (made up biochemistry but you get my point I hope  )

With that said, I think even todofwar's circuit level model could be useful. 

Kinases, for example have been identified as potential drug targets for treating a range of diseases. The problem with kinases is, that they're very similar (unsurprisingly since they're basically executing the same bit of chemistry on different substrates) and so developing a drug molecule that specifically inhibits a particular kinase is challenging. Chances are you'll get off-target effects, where your drug is moonlighting with other kinases and inhibiting them (to a greater or lesser extent) as well. Having an in-silico model of the whole kinome to study those effects could be tremendously useful. Maybe those off-target effects won't actually be that bad. Or - more excitingly - perhaps they're actually synergistic and a drug that targets kinases A, B and C is actually more effective than a selective drug against kinase A. You'd never rely entirely on the in-silico model of course but it could be another tool in the drug development box.

For that matter, having a model that would let you study the kinome as a single entity, rather than trying to dissect out a signalling pathway at a time, would be extremely interesting from a pure research point of view. Maybe this is already done though - I don't know.

Edit: More speculatively, could we compare our model with a real cell and use any discrepancies in observed behaviour of the two to identify new biochemistry. To use my example above, a model which only includes the interaction between kinase W and protein Y turns out to be flawed in various ways. Would it be possible to study those flaws, conclude that there's a new interaction at play, identify that interaction (between kinase X and protein Y) and then go looking for that experimentally?

Kind of like how crystal structures are solved (solve structure and refine against a model) but on a vastly more complicated scale.

 

TL:DR. We're not going to get a complete, high-fidelity, atomistic simulation of a cell. So can we build an abstracted simulation and would that be good enough to tell us anything useful?

Edited September 10, 2016 by KSK

[todofwar]

@KSK Spore was about evolving multicellular organisms, I was thinking the game map would be the body of a single cell. Think minecraft or factorio, setting up systems to feed resources to build more things. As it gets bigger, it requires more structural proteins, and at a certain point you unlock organelles maybe.