Quest to Native

[misiaczkowski]

For me good idea will be possibility to use "ghost puzzle" at share solutions. Few times I try to mix group solution with my solution but it's hard without ghost puzzle.

[infjamc]

I have several questions/comments:

1. Given that the goal of this puzzle is to get as close to the native structure as possible, why is it still scored the normal way rather than by minimum rmsd (root-mean-squared deviation)? For that matter, it seems to me that a low rmsd structure could still have a bad score.

2. Where are the loop modeling-specific tools? It would be really nice if one can cut open a loop, build from both sides, and then close the gap. (If anything, that's one of the methods actually used in computational protein structure prediction.) In this puzzle, for example, this technique could be used for the Gly #30-Trp #39 region.

[emcniel]

I 'm new to foldit and u until now I too only paid attention to the score and how to make it go up. I've just started this new puzzle and for the first time I could care less about the rank board. I'm focusing more on getting the shap right and as a result learning new ways to move the Protien. Thank for this one I think this is how we work the project past the casual gaming of moving my score up to realy focusing on the object of getting the truly best shape. I find this a great test for the project and a great learning chance for me as well.

[LennStar]

1. Because the RMSD is unknown (or would if we work on unknown proteins)

2. Protein design is planned…

[Steven Pletsch]

After some painstaking guided rebuilds, and a lot of time fiddling with the shape of the backbone, I have been able to match the shape of the backbone to the guide for the most part, however there seems to be some things that the software scores higher than the native configuration.

Hydrophilics seem to work best in the position show by the guide with the possible exception of Serine which I found difficult in some areas maintain the proper position during shakes.

Hydrophobics also seem to work best in the position shown by the guide for most of the hydrophobics, however here I found more exceptions than with the hydrophilics. For hydrophobics nearer the outside of the protein there were almost no exceptions in this puzzle, however for well buried ones or mostly buried ones, especially in the case of Tryptophan it is possible to get them to maintain the proper position, but there are other positions which score higher. Also with a couple sidechains, Proline and Cysteine being the biggest problems for me, it was difficult to maintain the proper position during shakes, though once worked, and wiggled out, the proper position scored very slightly higher. (~0.1 to 0.25 pts. higher)

From a learning perspective, I think this puzzle was invaluable, (The other one probably was too, but I did not get much time for it.) It showed me areas that I normally disregard, or pay little attention to but that can have significant impact on the puzzle. It also showed that I often focus on too small of an area and don't work larger ones enough.

Hope this helps in some way, I also hope to see a few more of these, as I think they are excellent for getting a feel for how things should be approached.

I would also like to propose that at some point we are able to work a puzzle which starts in the native configuration, and that we are tasked with raising the score as high as possible beyond it's native state, so that any discrepancies in scoring of models vs. the natural state might be possibly resolved in order to provide a more accurate scoring system. (I think the current one is great, but that there might be additional considerations which are not being entirely addressed, and by comparing native and higher scoring models it may shed some light on this discrepancy.)

[Steven Pletsch]

Also, I am curious which hydropathy index you are using, since looking into it, I see two different ones, and the amino acids which have given me the most problems seem to have the largest difference in values in the two indexes.

en.wikipedia.org

Hydrophobicity scales

Hydrophobicity scales are values that define the relative hydrophobicity or hydrophilicity of amino acid residues. The more positive the value, the more hydrophobic are the amino acids located in that region of the protein. These scales are commonly used to predict the transmembrane alpha-helices of membrane proteins. When consecutively measuring amino acids of a protein, changes in value indicate attraction of specific protein regions towards the hydrophobic region inside lipid bilayer. The hyd...

[vakobo]

If the purpose was to get maximaly close to native shape, why starting shape differs so much? It was very difficult to make it look like native shape.

[infjamc]

As we've seen from the results of the 2nd Quest to the Native puzzle, the native structure could have a much lower score than a near-native (10730 vs. 10829)… which means one of the following:

1. Bad score function; or
2. The native structure is NOT the global energy minimum in this case.

[Steven Pletsch]

One thing that is becoming very apparent in these puzzles is that the available rotamer positions to move the sidechain into sometimes exclude the one needed. even with many rebuilds and a lot of tweaking it is still sometimes impossible to achieve the proper rotation or shape

I don't know what could be done to change this, however if possible, it would be nice if in one of the modes, there was not restriction on the position.