>
> Next, a few comments on the model...
> The model could perhaps use some additional work. Note the Phe side chains
> exposed to solvent, where they prefer not to be (Phe 48, Phe 102).
Phe 48 is exposed to solvent, but phe 102 is quite acceptable. 102 is
places so it forms hydrophobic contacts with A94, A99 and L81. the
aromatic head isn't particularly exposed.
I calculated some statistics to give me a better idea of how exposed phes
are in crystal structures. Using iditis, I searched for all phes which
had their side-chains more than 10% exposed to solvent. The following
resulted:
Dataset: 1140 high resolution proteins, a total of 10767 phes.
Accessibility # of phes % of total phes
>10 4209 39.1
>25 2314 21.5
>50 774 7.2
>70 337 3.1
>90 107 0.99
So, from these stats, _some_ phes are totally exposed to solvent, but not
very many. Generally though, almost 40% of the phe sidechains do have
some exposure to solvent, albeit small.
Quanta has some nice routines for checking the 'health' of one's
protein eg conformer checks, exposed hydrophobics, holes due to
insufficient packing. Running the CheY model through these checks, I d
that whilst F48 was flagged as being outside the statistic norm for
exposure of that residue type, F102 was not.
> Note the
> helix which has its hydrophobic side exposed to solvent (Ile 88, Pro 89, Met
> 90, Val 92, and Ille 95).
Yes - there does seem to be a little problem with this part! Rotating the
helix by ~120 degrees will put the hydrophobic sidechains in contact with
other hydrophobes (I56, L64, F67 and L81). Q93 and Q96 should still be
exposed to the environment then.
> A minor point: Asp 87 at the start of that helix
> probably provides an H-bond to the main-chain N of Met 90, which is otherwise
> unsatisfied due to the fact that it is near the start of the helix.
This seems just to be an orientation problem (of the (Asp) C-O -> H-N
(Met)) which a little judicious tweaking of the structure will solve
(we're allowed to 'judiciously tweak' our structures, as they are models!).
_BUT_ - asp has a high propensity of being at the end of an n-terminal
a-helix (a 'helix breaker'), and as such may destabilize the helix at the
i-> i+4 th h-bond in order to break the helix structure (somehow - maybe
the interaction between D87 and H85 causes some strain and forces the
CO-HN interaction out of plane?). Otherwise, this part of the structure
seems reasonable.
One thing I've noticed (again running quanta over the structure), is that
there are some 'bad' (for bad read 'outside statistical norm'!) main
chain conformations, most notably the amide bond between T30-V31, which
is cis-, and not trans-. Unfortunately this happens a lot with quanta's
homology builder, which is sad. If you have quanta, it seems to be a good
idea to check the health of the protein after homology modelling, as the
health is not always acceptable (personally, I think that the homology
modelling package should work, expecially as we pay $000s for it...;)
Mike
All typos are deliberately introduced for confusion's sake... :)