Dear Joe
one thing that has helped me a lot with similar problems is G.
Kleywegt's SPASM - you make your best guess of the mainchain then add a total
guess of the sidechains (say with mutate in COOT - choosing the top rotamer).
Then you cut out the loop and run this in SPASM as a search model looking for
similar motifs in the pdb. You specify a low cut-off for the substitution
matrix - to get lots of hits - and use high values (3.0-4.5A) for the initial
positional errors for the residues - this means you get maybe 100 possible
structurally similar loops from other proteins. Then you can sort the output on
BLOSUM similarity score or on RMSD to your initial guess. The BLOSUM search
criterion is the reason that you add the sidechains. If you get no hits with CA
and SC (sidechain) then you can switch to CA only. Or try with a more
restricted bit of your loop. Or look through the output just for loops that
share a structurally conserved Gly or Pro with
your sequence.
If I get any possible hits I then superimpose them on the electron density to
see if they give hints how the sidechains might lie and also how the backbone
H-bonding might run. There are lots of small motifs with conserved sidechain to
mainchain H-bonding (there is a database of these in Glasgow I think) and so
you pick up hints from similar loops more often than you might think - from
completely unrelated proteins.
I have a script to run SPASM if you want it.
Another approach is RAPPER which builds multiple a priori loops very quickly
and sees if they fit the density. There used to be a RAPPER server... not sure
if it is still active.
Good luck.
Martyn
----- Original Message ----
From: Joe Smith <[EMAIL PROTECTED]>
To: CCP4BB@JISCMAIL.AC.UK
Sent: Wednesday, 15 October, 2008 4:11:01 PM
Subject: [ccp4bb] Poor electron density - polyAla or PolyGly?
Hello,
I have been building a protein model (resolution 2.2A) which has one
small loop of 6 residues having poor density. I cannot see any side
chain in this region but I can see relatively poor main-chain density
which at least clearly indicate the loop conformation. I am trying my
best to build polyAla chain in this region. But 3-4 residues end up
coming in disallowed region of Ramachandran plot. I think the problem
is - in this region i can not even build the main chain atoms with
high confidence.
Can I build polyGly in this region? or should I just leave this
region? I just don't feel like leaving this region empty.
Any suggestions in this regards is highly appreciated!
regards
Joe
