Coot Workshop

Part 2 - Building From Scratch

  1. Introduction
  2. Starting
  3. Baton Building
  4. Convert to Mainchain
  5. Assign Sequence
  6. Build Another Molecule
  7. Finishing Touches
1. Introduction

This tutorial is a slightly modified copy of this one from the Coot website. It assumes you are familiar with the concepts covered in part 1 of the workshop.

Due to advances in structure prediction (e.g. AlphaFold), building a model from scratch is very rarely needed. However, if structure prediction and automatic building both fail then building the molecule "by hand" may be the only way to proceed. It is also a good exercise to familiarise yourself with the shape of main-chain and side-chain density.

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 better than 20% in less than 30 minutes.

2. Starting

Open Coot, then click Calculate / 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 Display Manager until you have finished building, validating and refining. You should 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. Go to Calculate / Map Skeleton and change it to On.

3. Baton Building

We will now start to “baton build” the map. Go to Calculate / Other Modelling Tools to open another toolbar and click 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.

Baton View
Baton Controls

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 doubling back on yourself, which can be confusing the first time, but it is useful for extending existing chains.

Press Try Another and Previous Tip Position to cycle through the points, Lengthen and Shorten to change the baton length, and 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 (through Edit / Settings) then you can use the shortcut A (Shift-a) 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 Undo to move back one residue. The residue at the C-terminus is a cysteine that forms a disulphide bond. Click Dismiss on the baton controls when done. There may also be a molecule called Baton Atom Guide Points that needs deleting.

4. Convert to Mainchain

Now we need to turn these CA positions into mainchain. Use Ca Zone -> Mainchain in 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) so mainchain-backwards and Baton Atoms can be deleted.

The backbone now needs to be tidied up. Go to Calculate / All Molecule / Stepped Refine, pick mainchain-forwards and watch it as it goes. Is it making mistakes? That refinement may have gone to quickly to make a note of problem areas. Use Validate / Density fit analysis and real space refinement to correct anything that's wrong. Also use Edit / Change Chain IDs to label this chain as A.

5. Assign Sequence

One option for assigning the sequence is to use 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 assign the sequence for us. Go to Calculate / Assign Sequence / Assign Sequence (py) to get the following window:

Sequencing

Select mainchain-forwards to be sequenced, turn on Auto-fit-refine after sequencing?, change Chain ID to A then copy and paste the following sequence: 

DVSGTVCLSA LPPEATDTLN LIASDGPFPY SQDGVVFQNR ESVLPTQSYG
YYHEYTVITP GARTRGTRRI ICGEATQEDY YTGDHYATFS LIDQTC
Click 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 View to look for alanine residues that don't match the sequence (possibly at the termini) and mutate them into the correct residues (or delete them if they are extra).

6. Build Another Molecule

Now we need to build another molecule (the NCS related copy). We will first just build a small part. Search around to find a helix not already built (and not a crystallographic symmetry copy of the chain already built). This may be easier using the map skeleton with the side chains pruned. Then use Place Helix Here in Other Modelling Tools to place a helix in the density. Make changes to the helix so that it fits the density well, then assign the sequence (and residue numbers) and fit the side chains using the Assign Sequence (py) tool again.

Now we will superpose a copy of the A chain onto this helix. Go to Calculate / LSQ Superpose:

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, first click 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 mainchain-forwards molecule.

Instead of building part of the next chain and superposing, it is also possible to find and fit a copy of a chain using map masking and Jiggle-Fit tools as shown in the following video:

After the new chain is merged, Coot should detect that the A and B chains are NCS copies of each other. Now you can use the shortcut o 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 under Calculate / NCS Tools.

7. Finishing Touches

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, paul.bond@york.ac.uk