Error computing BuriedUnsatHbonds

[CarlosChacon-cell]

Hi!,
I am trying to compute the buried unsatisfied Hbonds as the people from BindCraft does using the PyRosetta suite.
However, whenever I try to run it I get the following error:

Error! You must set -holes:dalphaball to use dalphaball_sasa!!

The problematic line is:

  buns_filter=pyrosetta.rosetta.protocols.rosetta_scripts.XmlObjects.static_get_filter('<BuriedUnsatHbonds report_all_heavy_atom_unsats="true" scorefxn="scorefxn" ignore_surface_res="false" use_ddG_style="true" dalphaball_sasa="1" probe_radius="1.1" burial_cutoff_apo="0.2" confidence="0" />')

It does not raise an error if dalphaball_sasa is set to 0.

I found a discussion that shows how to solve it, but I wonder if it is neccessary or if dalphaball_sasa=0 is still good ? I have not found any documentation detailing this

Thanks in advance!

[roccomoretti]

If dalphaball_sasa=1 is not set (or if dalphaball_sasa is explicitly set to false), an alternative method of calculating solvent accessible surface area will be used. (That is, it is not like the SASA calculation is being skipped.) Since they're both attempting to calculate the same thing, the values you get will be mostly the same. That is, there will be a high degree of correlation between the two calculations. However, there is likely to be slight variation in details, which may or may not affect a particular structure. As such, the exact number of Buried Unsatisfied Hydrogen bonds you calculate possibly will be different between the two calculations. (I would probably trust the dalphaball method to be the "more accurate" one with respect to SASA, but buns determination is an inexact science -- I'm not sure how much the difference in SASA calculation accuracy affects how well the buns metric correlates with protein quality.)

That's mainly going to express itself for those hydrogens bonders being on the edge of being buried or not. Clearly buried unsatisfied hydrogen bonders or those which are clearly solvent exposed are unlikely to differ between the two.

How much difference that's going to make is likely going to be up to how critical the BUNS calculation is to your protocol. If that's a somewhat loose threshold, it may not matter much, as you're probably already including things which are on the bad side of marginal anyway. It's likely to be more of an issue if the BUNS filter is quite strict, as small variations in calling any particular unsatisfied hydrogen bonder have a much greater effect in whether you accept or reject a design.

In the particular case of BindCraft, I don't know if anyone has taken a look at the effect of different BUNS calling methods. You might be able to ask the authors. Alternatively, I'd probably just recommend trying it -- if you get designs which look reasonable you're probably fine. You may need to play around with the BUNS thresholds, loosening or tightening them in order to get a sufficient number of designs which look decent.

[CarlosChacon-cell]

Thanks!