When automatic building fails, typically because the resolution limit of your data is too low, then building the molecule "by hand" may be the only way to proceed. Recognizing the shape of main-chain and side-chain densities is valuable and this tutorial aims to introduce these to you. Note that this tutorial map is an easy map to build into, the side chains are (mostly) clear. Using just a map and a sequence, we will attempt to generate a model. This model can then be validated and refined with Refmac for several rounds. With some experience you should be able to get an R-factor of less than 20% in less than 30 minutes.
Coot, then click
Extensions / Load tutorial model and data
to load the structure.
It has two chains with 96 residues in each.
Start by going to the C-alpha (CA) of residue 2 in the A chain.
The idea is to build the complete A chain in the forward direction.
Make a note of the position of the C-alpha in residue 1 then
hide the molecule in the
until you have finished building, validating and refining.
It is a good idea to also hide the difference map
because we are not interested in the previously built model.
The backbone trace through the density
can be more easily visualised
by using a skeleton map.
Calculate / Map Skeleton
and change it to
We will now start to “baton build” the map.
Calculate / Other Modelling Tools to open another toolbar
C-alpha Baton Mode.
A set of controls will appear,
as well as a white baton
and a set of pink points that show possible places for a C-alpha
that are 3.8 Å away from the current position.
Remember, when you start, you are placing a CA at the baton tip for residue 1 (the N-terminus). After placing CA for residue 1, you will get a choice of positions for residue 2, which is currently at the centre of the view. This might seem that you are "double-backing" on yourself, which can be confusing the first time, but it is useful for extending existing chains.
Try Another and
Previous Tip Position
to cycle through the points,
Shorten to change the baton length,
Accept to place a CA and move onto the next residue.
If none of the guide points are suitable
you can use
b to toggle baton swivel mode.
If you have installed the template keybindings
Extensions / Settings)
then you can use the shortcut
to accept the baton position.
Build from the N-terminus to the C-terminus.
There are 96 residues to build and it takes about 15 minutes or so.
If you make a mistake you can press
to move back one residue.
The residue at the C-terminus is a cysteine that forms a disulphide bond.
Dismiss on the baton controls when done.
There may also be a molecule called
Baton Atom Guide Points
that needs deleting.
Now we need to turn these CA positions into mainchain.
Ca Zone -> Mainchain
Other Modelling Tools
and click one of the baton atoms.
You will need to wait several seconds while it builds
(note that you need at least 6 residues for this to work).
Two new molecules will be created that are traced in different directions.
We know the forward direction is correct
(see how much better the carbonyls fit)
can be deleted.
The backbone now needs to be tidied up.
Extensions / All Molecule / Stepped Refine,
and watch it as it goes.
Is it making mistakes?
That refinement may have gone to quickly to make a note of problem areas.
Density fit analysis
and real space refinement to correct anything that's wrong.
Edit / Change Chain IDs to label this chain as A.
One option for assigning the sequence is
Mutate Residue Range
but this is only useful if you know the identity for all the residues.
If we know the sequence for the whole chain
but have only built part of it
then we can also ask Coot to dock the sequence for us.
Extensions / Dock Sequence / Dock Sequence (py)
to get the following window:
mainchain-forward to be sequenced,
Auto-fit-refine after sequencing?,
Chain ID to
then copy and paste the following sequence:
DVSGTVCLSA LPPEATDTLN LIASDGPFPY SQDGVVFQNR ESVLPTQSYG
YYHEYTVITP GARTRGTRRI ICGEATQEDY YTGDHYATFS LIDQTC
Sequence closest fragment. Coot will think for several seconds while assigning the sequence before trying to fit the side chains. The sequencing may fail depending on how good the model is. You need to make sure the CBs are in density and pointing in the right direction and the fragment is long enough. Use
Draw / Sequence Viewto look for alanine residues that don't match the sequence (possibly at the termini) and mutate them into the correct residues.
Now we need to build another molecule (the NCS related copy).
We will first just build a small part.
Using the map skeleton, search around to find a helix
not already built (and not symmetry related to the model already built).
Place Helix Here
Other Modelling Tools
to place a helix in the density.
Make changes to the helix so that it fits the density well,
then dock the sequence and fit the side chains.
Now we will superpose a copy of the A chain onto this helix.
Calculate / LSQ Superpose:
Make sure you are moving a copy of chain A onto the new helix.
A new molecule should appear that almost fits the density.
To clean up the fit with a rigid body refinement,
Rigid Body Fit Zone in the refinement toolbar
then click the N-terminus and C-terminus.
Then perform a stepped refinement of the whole molecule
and merge the the copy back into the original.
Unfortunately, there is no slick way to make Coot rebuild ghosts
for this composite molecule that contains both the A and B chains.
We need to save the coordinates and open them again.
Check the console as you open the molecule,
Coot will tell you that there are NCS related molecules.
Now you can use the shortcut
to quickly switch the view between NCS copies.
Also, if you make changes to one chain and want them copied into other chains
you can do this using
Copy NCS Residue Range and
Copy NCS Chain
Extensions / NCS.
If you want to finish the structure you've built from scratch, go through rounds of validation and model building using Coot and refinement using REFMAC in CCP4i2. Then waters need to be built as well as a ligand (3GP).
Paul Bond, University of York, firstname.lastname@example.org