[Molvis-list] Will the real hbonds please stand up?

Eric Martz emartz at microbio.umass.edu
Sun Mar 23 18:19:49 EDT 2008

On 3/23/08, Frieda Reichsman asked
(about FirstGlance in Jmol [FGiJ]'s Contacts display):

>Hi Eric,
>In a [protein chain] dimer, if I click on either chain and look at 
>putatively hbonded
>non-water, shouldn't I see the same number of contacts from one
>protein chain to the other, regardless of which is the target?
>In 1LLI, I see 6 for chain A and 8 for chain B.
>In 1YSA, I see 3 for chain C and 5 for chain D.
>Maybe I am missing something about hbonds, but this doesn't make sense
>to me right now.

[Here's how to show these putative hbonds in FGiJ:
  - Display 1YSA in FGiJ
  - click "Contacts"
  - click on one of the protein chains (it is now the "target")
  - click "Show Atoms Contacting Target"

To limit the display to the putatively hbonded non-water:
  - click "Uncheck"
  - check "Show putatively hydrogen bonded non-water"
  - click the 4th thumbnail display mode (all balls and sticks)]

You are right: If FGiJ could really determine where hydrogen bonds 
are, in most cases there would be an equal number of donors + 
acceptors on each chain.

The problem is the crudeness of FGiJ's reporting of "putative hbonds".

As you know, FGiJ (and Protein Explorer) designate any oxygens or 
nitrogens within 3.5 A of other oxygens or nitrogens as "putatively 
hbonded". They don't know where the hydrogens are, and don't take 
angles into account. So no donor or acceptor assignments are made, 
just proximities of possible donor/acceptor pairs.

Take your second case, 1YSA.

There are two simple atom pairs:
   Glu270:C.OE2 - Lys275:D.NZ. 3.1A
   Asn264:C.OD1 - Asn264:D.ND2 2.6A
Each puts one "putatively hbonded" atom in each chain. So far, no problem.

The third interaction is ambiguous:
   Asn264:D.OD1 is close enough to three possible partners:
     Asn264:C.N  2.8A
     Lys263:C.NZ 3.1A
     Leu260:C.O  3.2A
That is why you get two more "putatively hbonded" atoms in :C than in :D.
Which one of these is really hbonded? (Or might two be sharing one hbond?)

A next step could be to add hydrogens to the PDB file model with 
Molprobity, and let Molprobity flip the Asn264's in both chains or 
not as it sees fit. Then one could try to pick out donors and 
acceptors. However, analysis of the relevant angles is something I 
don't know how to do.

The resolution of 1YSA is 2.9A, perhaps a bit low to be trying to 
make precise assignments. The temperature factors of the above-listed 
atoms are relatively low, so that isn't an additional problem (but 
its always important to check). (In FGiJ, while showing these 
contacts, click "More Views.." and then "Color by Uncertainty".)

Hope that helps, -Eric

More information about the Molvis-list mailing list