The details do matter

Consider a protein-ligand binding model. How easy is it to predict the best and worst binders in terms of affinity? Now, how hard is it to quantitatively rank these binders in terms of free energy of binding?

The former, while not an easy problem, has been solved in various ways multiple times for individual problems. In fact a new docking program is expected to at least achieve the minimal goal of ranking the most active ligands at the top and the least active at the bottom.

However, in spite of impressive advances, the latter problem is still regarded as a holy grail.

Now consider molecular dynamics simulations of proteins. Coarse-grained MD approximates atomistic details by subsuming them into a broader framework; for instance, "united atom" force fields will sometimes treat the hydrogen atoms attached to carbons implicitly without explicitly representing them. Coarse-grained MD has been indispensable for simulating large systems where explicit representation of fine details would be prohibitively time-consuming. But coarse grained MD would not always be able to shed light on cases where the fine details do matter, such as proton transfer in enzyme active sites and the general detailed modeling of enzymatic reactions.

Finally, consider solvation models in molecular simulations, a topic of perpetual development and high interest. Implicit models where the solvent and solute are considered as mean dielectrics and their interaction is modeled as a sum of electrostatic and non-electrostatic interactions are all the rage. They frequently work very well and I have myself used them numerous times. But consider cases where the detailed thermodynamics of individual water molecules in protein active sites need to be modeled. Implicit solvation can be of scant use in such circumstances. The use of implicit solvation often makes general predictions about qualitative differences between protein-ligand interactions possible, but it can mask the detailed reasons for those differences and indeed cannot even account for such differences many times.

Something similar seems to be happening for climate change. It is relatively easy to make general statements about extreme events occurring. It is generally true that putting all that buried CO2 back into the atmosphere as a high entropy substance is probably a bad idea, and that cutting emissions is probably a good idea. My problem is not so much about politicians suggesting such general solutions as it is about them sounding crystal clear about all the scientific details. It's much harder to predict the details about individual effects and 'rank' them in terms of their severity, nor is it easier to rank individual solutions to the problem in terms of relative impact. That is something that is an inherent limitation of the science at this point, and any good scientist worth his salt should acknowledge this. Nonetheless, the science has been declared 'settled' and politicians seem to suggest implementing concrete policies in spite of the coarse-grained nature of the problem. As I mentioned before, the fathers of empirical inquiry Newton, Bacon, Locke, Boyle and Hume would have been rather chagrined with this state of affairs.

It's even more disconcerting to realize that activists propose solar and wind power which could be useful in limited amounts but have by no means proven to be robust, as global high energy density solutions. On the other hand there is nuclear power, a proven existing technology that packs more energy than any other, is clean, decidedly CO2-free and highly efficient and deployable. Yet the same politicians who condemn fossil fuels and talk about climate change constantly fail to tout the one solution that could solve the problem they are trying to address. What can you say when someone ignores a solution to a problem that's staring them in the face?

Freeman Dyson, who has been duly and gratuitously vilified for his skepticism about climate models, said that he lost interest in climate change when the issue turned from scientific to political. One can understand why he said that. The exemplar of tentative scientific understanding was Niels Bohr, and Einstein's quip about him captures the perfect attitude we should all have about complex scientific issues, an attitude that is sadly lost on many climate modelers; Einstein said of Bohr that "He looks like someone who never behaves as if he is in possession of the truth, but one who is perpetually groping".

Gropers are especially encouraged to apply to The Academy.

2 comments:

  1. This is true that Niels Bohr has a good point.

    ReplyDelete
  2. All good points about solvation & thermodynamics, but these may be more obfuscatory than enlightening when it comes to molecular modeling. They may also be computationally unattainable. Perhaps if one compared very similar molecules with different activities (i.e. agonist vs antagonist or substrate vs inhibitor), then the essential differences would be more noticeable. We began to do this - see: http://cogprints.org/4095/ but no one seems to have followed up with it.

    ReplyDelete

Markup Key:
- <b>bold</b> = bold
- <i>italic</i> = italic
- <a href="http://www.fieldofscience.com/">FoS</a> = FoS