Pyrosetta energy function

[Anonymous]

I am considering various conformations of a given protein and I should compare their corresponding energy values. I have used Pyrosetta to compute the energies using the “scorefxn = create_score_function(‘standard’)” command. I was under the impression that the PDB structure of a protein should have the minimum energy.
However, during my computations I have cases in which the alternative conformations have lower energies according to Pyrosetta. Here is a simple example: I considered the protein with PDB ID 1CQW and one of its alternative conformations. Pyrosetta gives energies -487.858056738 and -488.060045834 for the original and alternative structures, respectively. So my question is whether this is possible, i.e., can an alternative conformation of a protein have energy lower than its PDB structure.

Thanks,
–Somayyeh

[Anonymous]

There is nothing wrong here.

First, physics. The crystal structure is the lowest energy structure if it has been solved to infinitesimal resolution with no errors. This has never happened and probably never will.

Also, a crystal structure is a static (and usually mostly desolvated) image, but actual proteins are dynamic and solvated, so they will occupy an ensemble of structures in any real situation. We assume the crystal is representative of that ensemble, but it might not be for many reasons (crystal packing, most importantly).

Second, Rosetta’s scorefunction is not a perfect match for actual physics, it has errors. So whatever Rosetta thinks is best is likely to be at least slightly different from what real physics thinks is best.

In fact, most straight-from-the-PDB structures have horrible Rosetta energies; we relax them into Rosetta’s energy function before using them. This generally gives them a small RMSD from the crystal structure, but makes their energies reasonable. Both the crystal structure and the Rosetta-relaxed structure are likely to be representative of the real physical ensemble, so, it’s good enough.

Finally, the two energies you quote are trivially different. On the protein scale energy differences like that (one energy unit) are meaningless.

[Anonymous]

Rosetta’s energy function has evolved over the years, and unfortunately there is no one publication I can point you to that will explain how each of the scorefunction terms works and how they relate to one another. Start with the recent reviews and track backwards is the best advice I can give. Rohl et al., 2004 (Methods in Enzymology) along with the supplementary info for the Top7 paper (Kuhlman B, Dantas G, Ireton GC, Varani G, Stoddard BL, Baker D (2003-11-21). “Design of a novel globular protein fold with atomic-level accuracy”. Science 302 (5649): 1364–1368. doi:10.1126/science.1089427. PMID 14631033.) are good places to start.

[Anonymous]

PDFs from a workshop at the Meiler lab give some decent overviews: http://www.meilerlab.org/index.php/jobs/resources [Rosetta Tutorials]

[Anonymous]

Thanks for your useful replies. I have another relevant question. As I said, I need to consider alternative conformations of a given protein structure. Based on the energies of these conformations I want to decide whether a conformation is acceptable or not. A conformation C is acceptable if the probability of transition from the initial conformation (PDB structure) to conformation C is reasonable (the energy of conformation C is not much higher than the energy of initial conformation).

I use PyRosetta to compute the energies of conformations. My question is: what is an acceptable energy difference? In other words, what is a good choice for parameter alpha in the following statement: An alternative conformation C is acceptable if its energy is not more than alpha units higher than the energy of initial conformation.

Thanks,

[Anonymous]

Depending on how you are doing your sampling, your ensemble of possible conformations will be a thermodynamic ensemble and you could look at them using Boltzmann statistics – X many units of energy means Y times less likely a conformation.

Less rigorously, look for a natural cutoff in your models. You may find that there is a steep change in energy after the top 3% of models, so only consider the top 3%. Or, just take any models that don’t have an identifiable flaw and are within your resources to spend time considering.

This question is best answered empirically; the right way to do it is whatever works.

[Anonymous]

Thanks for your prompt and complete response. I have another relevant question. I am looking for a detailed description of Pyrosetta scoring function.

Thanks,
–Somayyeh

[Author]

Posts