Talk:Macromolecular docking

Field of what

If anyone can provide an example of Protein-protein docking being performed by a theoretical or quantum chemistry group, then that could be reinstated.. otherwise I think it is misleading. Zargulon 09:36, 5 September 2005 (UTC)

I am not an expert in this field but I cannot imagine how theoretical biochemistry could not be considered as a subfield of theoretical chemistry or quantum chemistry. I guess the people doing protein-protein docking are using density functional theory, molecular mechanics or molecular dynamics, aren't they? Are the methods used in protein-protein docking ignoring quantum mechanics? Theoretical chemistry is defined as any non experimental reasoning applied to chemistry not as what are doing theoretical (or quantal) chemistry groups. Quantum chemistry is defined by the application of quantum mechanics to chemistry whoever does this. Am I wrong? --09:57, 5 September 2005 (UTC)

Density functional theory is not used. Molecular dynamics is used only by a minority and with negligible success. Protein-protein dockers are happy with heuristic solutions which use, for instance, residue contacts, or shape complementarity of molecular surfaces. Theoretical chemistry approaches are generally too time consuming to be useful in the case of even a small protein. There is some limited cross-fertilization of ideas but they are different communities and address different questions. Zargulon 23:53, 5 September 2005 (UTC)

OK Thanks. I understand better. Do those heuristic solutions have something to do with molecular mechanics? I nevertheless think what they do is somehow theoretical chemistry. Because it is a non experimental reasoning.Vb13:51, 6 September 2005 (UTC)~

Yes, some heuristic methods occasionally use parameters from molecular mechanics force fields (like CHARMM or AMBER). It is a matter of finding something that works! Zargulon 18:06, 6 September 2005 (UTC)

OK I have re-edited to link with theoretical chemistry but not with quantum chemistry. I guess you now agree --131.220.68.177 08:56, 7 September 2005 (UTC)

I don't agree with the first sentence, but I agree that there should be a link to theoretical chemistry, which I am retaining. the point about CHARMM and AMBER is that protein-protein dockers hijack the parameter sets from these codes, not that they actually run them. Zargulon 10:01, 7 September 2005 (UTC)

Also I am removing the link to computational chemistry.. it sounds like it is a good idea, but it becomes clear from looking at the introduction to the computational chemistry page that there is absolutely no overlap with protein-protein docking at all! Zargulon 10:01, 7 September 2005 (UTC)

Merge

Disagree Zargulon 11:34, 6 January 2006 (UTC)

Disagree. I am against merging this article with the one on Protein-protein interactions. Protein-protein docking is a bioinformatics method. Protein-protein interactions are an observable.--Miguel Andrade 15:20, 24 January 2006 (UTC)

Thanks for the clarification. I've removed the merge tags. However, I am wondering why there is no wikilink or direct mention of protein-protein interactions, but only of interaction. --Uthbrian (talk) 18:28, 24 January 2006 (UTC)

Edits August 9

• Primary structure prediction implies quaternary structure prediction. The reverse is not true. What is called "docking" (for macromolecules) always produces a set of coordinates thus predicting primary structure. It is therefore more precise use primary structure in the lead. If it only predicts quaternary structure and does not produce a full coordinate set it is called something else, e.g. "binding site prediction/modelling" or "prediction/modelling of the mode of interaction".

• The interactome is the set of all protein-protein interactions in a genome (or, occasionally, more than one genome). It does not refer to the set of all protein-protein interactions of an arbitrary protein or set of proteins. Zargulon (talk) 22:50, 9 September 2008 (UTC)

• All structure prediction is modelling. Not all modelling is prediction. Not all docking is structure prediction; firstly, a lot of docking involves making models which satisfy some optimality criterion but could never possibly exist (so not prediction), e.g. bound-bound docking and other scenarios. Secondly in cases where the complex structure is known e.g. the blind trials and all benchmarking scenarios, this is not prediction, it is building models and seeing how good they are by comparing them with an already known correct answer. Zargulon (talk) 22:57, 9 September 2008 (UTC)

• Primary structure prediction implies quaternary structure prediction – The primary structure (the amino acid sequence) is NOT what is predicted by macromolecular docking. A prerequisite for docking is knowledge of the primary/secondary/tertiary structure.

::* A prerequisite for docking is (usually, but not always, see homology modelled targets) the primary structure of the interactors. The aim is modelling the primary structure of the complex. Zargulon (talk) 08:38, 10 September 2008 (UTC)

• What is being predicted is the quaternary structure. For example, in reference #3:

Prior to the successful determination of the structure of the complex, nine laboratory groups were given the refined atomic coordinates of each of the native molecules ... All six of the groups that accepted the challenge correctly predicted the general mode of association of BLIP and TEM-1.

::* What is being modelled (not predicted, despite the word used in the paper) in that experiment, and all docking, is the primary structure. The quaternary structure was, as you point out, successfully modelled. The primary structure, less so. Zargulon (talk) 08:38, 10 September 2008 (UTC)

• The refined atomic coordinates define the primary, secondary, and tertiary structure. The general mode of association is the quaternary structure. In rigid docking, the tertiary structure remains unchanged. In flexible docking, parts of the tertiary and perhaps even the secondary structure could change, but the primary structure would never change.

::*Flexible docking involves a conformer search of at least a subset of sidechains, if not the backbone, of the interactors. This by definition changes the coordinate set, and thus the primary structure. Even rigid body docking outputs coordinate models with precise atomic coordinate information which is not present in the structure of the individual interactors, so this may be said to model the primary structure of the complex. Zargulon (talk) 08:38, 10 September 2008 (UTC)

• It is therefore more precise use primary structure in the lead. – Using the term primary structure to describe what is predicted by docking is incorrect. It is the quaternary, not primary structure which is being predicted. This follows directly from the definitions of primary and quaternary structure.

::*See above.Zargulon (talk) 08:38, 10 September 2008 (UTC)

• The ultimate purpose of modeling is prediction. In order to make a prediction, it may be necessary to construct and evaluate many thermodynamically unstable configurations of the complex. In the lead of an article which is meant to be understood by a wide audience, is is much clearer to use the term prediction since it much more precisely describes what is ultimately trying to be accomplished by the modeling. Cheers. Boghog2 (talk) 06:21, 10 September 2008 (UTC)

• The most important thing about the lead, as well as the rest of the article, is that it be correct. Clarity comes a close second. People in the field aim to answer a number of questions (not just structural), most of which can be said to lead, perhaps indirectly, to the ability to predict something. However, docking is often modelling rather than prediction. It is important not to confuse the thing that is done with the reasons why people do it. Zargulon (talk) 08:38, 10 September 2008 (UTC)

Sorry.. I have indeed misunderstood primary structure, you're right that it refers to sequence. Docking models what is on Wikipedia called molecular geometry. Zargulon (talk) 08:44, 10 September 2008 (UTC)