Improve MLP lipophilicity parameters

[zhijie.li@…]

From: Zhijie Li
Thu Jul 11 21:48:09 PDT 2019
Previous message: [chimerax-users] WebGL apps still in the pipeline?
Next message: [chimerax-users] MLP value of the ASP and GLU carboxylate
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Hello,

I noticed that the lipophilic atomic potentials of OD2 and OE2 are set to very positive values in mlp.py:

'ASP': {'C': -0.54,

'CA': 0.02,
'CB': 0.45,
'CG': 0.54,
'N': -0.44,
'O': -0.68,
'OD1': -0.68,
'OD2': 0.53}

'GLU': {'C': -0.54,

'CA': 0.02,
'CB': 0.45,
'CD': -0.54,
'CG': 0.45,
'N': -0.44,
'O': -0.68,
'OE1': -0.68,
'OE2': 0.53},

As as result the hydrophobic surface representation always show hydrophobic patchs on the carboxylates, which look quite odd to me. Is there a justification for this setting? In addition, aren't the two oxygens in the carboxylate equal considerting that they are in resonance?

Zhijie

[Tom Goddard]

Elaine made new parameters as she describes below:

Hi All,
Due to some limitations of the original set of atomic lipophilicity values from pyMLP used in our “mlp” command, most noticeably asymmetry and sign difference in the values for ASP/GLU carboxylate oxygens as pointed out by Zhijie on our mailing list, I investigated instead using the atomic lipophilicity values in this paper:

Prediction of Hydrophobic (Lipophilic) Properties of Small Organic Molecules Using Fragmental Methods:  An Analysis of ALOGP and CLOGP Methods. Ghose AK, Viswanadhan VN, Wendoloski JJ.

10. Phys. Chem. A1998; 102(21):3762-3772.

<​https://pubs.acs.org/doi/full/10.1021/jp980230o>

These are the values used by the PLATINUM webserver for molecular lipophilicity surface coloring.
<​http://model.nmr.ru/platinum/>

This paper lists very many atom types, much more subdivided than in ChimeraX, and include about a dozen different types of explicit hydrogens. It's not feasible for us to encode the chemical rules to recognize all of these types currently, and we want to allow MLP coloring without requiring hydrogen addition, so I manually created a lookup table with the same amino acids as the original pyMLP set in which the values for the (inferred) attached hydrogens were added to the respective heavy-atom values. To this list I added a few more residue types that could occur within a protein chain (our “mlp” command ignores nonprotein): MLE and the peptide-capping residues NH2, NME, and ACE, as well as UNK (backbone only, sometimes used for lower-resolution structures in which the amino acid type cannot be determined from the density).

I call that set “Ghose-united.” I also made a variant set “Ghose-united-shifted” in which 0.03 was added to each atomic value (including any virtual hydrogens, so some atoms got 0.03 but others could get up to 0.12 added). This ad hoc shift was recommended but optional at the PLATINUM webserver.

I didn’t go to all this effort for Zhijie alone… the parameters had been bugging me for a while, and he’s right that the icon button makes this ChimeraX feature very prominent and we want it to give reasonable results.

Much more detail/images on testing here, so I’ll skip to the punchline. (ONE QUESTION: is it OK that the mlp.py variant files are publicly available as links from this page?)
<​http://www.rbvi.ucsf.edu/chimera/data/mlp-sep2016/#update>

CONCLUSIONS:
I recommend using the Ghose-united parameters as the new default and keeping the default coloring range the same. Maybe the original atomic lipophilicities could be offered as an option, but I don’t think we need to offer Ghose-united-shifted. If we don’t want to bother with options, I’m happy enough with them that Ghose-united could be the only choice.

Quite consistently on a variety of proteins, compared to the original parameters, the Ghose-united parameters shift the minimum surface value from the negative forties to the negative twenties and the mean closer to zero, but give about the same maximum. Coloring over the same default range (-20,20) gives the same overall impression, with the same regions coming out as hydrophobic, and not surprisingly, a more reasonable result for Asp/Glu sidechains and UNK/bb-only residues. The additional ad hoc shift (ghose-united-shifted parameters) doesn't seem to add anything to the analysis. Anyone who prefers a more saturated coloring could use a more restricted coloring range (e.g. -15,15) but I think the -20,20 range provides a better sense of the range of values anyway.

Below I attached the three variants of mlp.py:

mlp-orig.py - the only difference from mlp.py in current distrib is that I fixed wrong atom names in the HYP residue
mlp-ghose-united.py
mlp-ghose-united-shifted

… and images for just two of the examples from my detailed writeup. Perhaps the bicolored UNK chain in the third image is a bug… should be all white since its min,mean,max are all zero.
<​http://www.rbvi.ucsf.edu/chimera/data/mlp-sep2016/#update>

[Tom Goddard]

Changed ChimeraX mlp command to use Elaine's Ghose united parameters.