It's hard to answer whether your R/Rfree is "good enough" for publication, because "good enough" is a value judgment and not a concrete yes/no decision.

First, I assume that you have examined the electron density at every residue, water, ligand, etc., *manually* and are not simply relying on validation parameters in Coot or Molprobity to detect troublesome features? Automatic detection of problems in the structure is very helpful, but it almost never detects every single error or problem.

If that's true, then there are a few things you can take a look at:

1. Most importantly, what is the resolution of your data? An R/Rfree of 25 and 31% would generally be very unsatisfactory if you're working at high resolution (better than ~1.5 Å), but those same R/Rfree values would in fact be quite reasonable if you're working at modest or low resolution, say worse than 2.5 Å. We need a little more information to get an accurate idea of the expected quality of your structure.

2. Have you added a reasonable number of clearly identifiable waters (first and second hydration shells, etc) to your structure and deleted any spurious waters that are not supported by density and/or do not have clear interactions with other waters or protein residues?

3. If you're working at high resolution, have you accurately modeled alternate side chain conformations?

4. Are you modeling B-factors appropriately - anisotropic at very high resolution and isotropic otherwise?

5. If you have any co-factors or ligands, etc., have you refined their occupancies, and did the occupancies refine to self-consistent values for covalently bonded atoms?

6. Are there any unmodeled blobs of density in the Fo-Fc difference map that need to be explained?

These are just some of the places you can look to explain R values that are "too high." When you refine again, make sure that R and Rfree are either both decreasing, or that the gap between them is decreasing. What you want to avoid is any refinement that leads to a reduction in R but no reduction in Rfree, because that is an indication that you might be over-fitting the data.

Good luck!

It's hard to answer whether your R/Rfree is "good enough" for publication, because "good enough" is a value judgment and not a concrete yes/no decision.

First, I assume that you have examined the electron density at every residue, water, ligand, etc., *manually* and are not simply relying on validation parameters in Coot or Molprobity to detect troublesome features? Automatic detection of problems in the structure is very helpful, but it almost never detects every single error or problem.

If that's true, then there are a few things you can take a look at:

1. Most importantly, what is the resolution of your data? An R/Rfree of 25 and 31% would generally be very unsatisfactory if you're working at high resolution (better than ~1.5 Å), but those same R/Rfree values would in fact be quite reasonable if you're working at modest or low resolution, say worse than 2.5 Å. We need a little more information to get an accurate idea of the expected quality of your structure.

2. Have you added a reasonable number of clearly identifiable waters (first and second hydration shells, etc) to your structure and deleted any spurious waters that are not supported by density and/or do not have clear interactions with other waters or protein residues?

3. If you're working at high resolution, have you accurately modeled alternate side chain conformations?

4. Are you modeling B-factors appropriately - anisotropic at very high resolution and isotropic otherwise?

5. If you have any co-factors or ligands, etc., have you refined their occupancies, and did the occupancies refine to self-consistent values for covalently bonded atoms?

6. Are there any unmodeled blobs of density in the Fo-Fc difference map that need to be explained?

These are just some of the places you can look to explain R values that are "too high." When you refine again, make sure that R and Rfree are either both decreasing, or that the gap between them is decreasing. What you want to avoid is any refinement that leads to a reduction in R but no reduction in Rfree, because that is an indication that you might be over-fitting the data.

Good luck!

Matthew Whitley,

Thanks for replying my question. The suggestions you made are very elaborate, i first of all appreciate your patience in writing the reply.

Coming to the resolution i am working with a ~2.2 A structure.

The answer to the point #2 that you have raised is not completely resolved, i can see only few water moles which i can identify as waters. But, i deleted some of the spurious waters.

Point# 3 is a good suggestion , but i don't see alternative conformations ( in some cases yes) with the resolution i am working with.

Point#4 i did not resolve completely, probably i will work on it right away.

I have addressed point # 5 and 6 already with my most recent refinement.

Finally i completely agree with your concluding remark of making less gap between R and R free.

I shall work on it and try my best to further refine the model.

Thanks Again,

I will keep you posted if i make more significant progress.