Building macromolecular structures at high resolution with COOT

CHEM213A Winter 2022 Homework #2

For this exercise, you’ll first need to get COOT running on your machine. Please follow the
installation instructions in the separate “Coot installation instructions” document.

The goal of this exercise is to gain first-hand experience with building macromolecular models at
high resolution, and to learn how proteins assemble in three dimensions. You should start to
see how hydrophobic interactions, hydrogen bonds, and secondary structures work together to
provide the folding energy for a protein.

You will be provided two files:

hw2_electron_density.mtz – A MTZ-format file with coefficients for 2FoFc and FoFc maps.
hw2_starting_model.pdb – a PDB-format coordinate file with the backbone (but not side
chains) of a protein.

Your task is to complete the model of this protein and submit:

(1) the final coordinates in PDB format;
(2) answers to the questions on the next page.

Both the model and the answers are required for full credit on this assignment.

Building the model:

1) First, open COOT. Open the model PDB file using “File > Open coordinates.” Open the
MTZ file containing map coefficients using “File > Auto Open MTZ”. You’ll need to build
into the 2Fo-Fc map (blue). A quick guide to the COOT interface is provided in a
separate PDF file.
2) You have the backbone (main chain, but no side chains) for residues 75-280 of this
protein. Your first task is to complete the backbone trace by building residues 281-326.
Use the “Go To Atom…” dialog box to center on residue 280, then click the “Add
Residue…” button and click on 280. It will add residue 281 (as an alanine) where it
thinks it should go. If you agree, continue on to residue 282. If not, you may need to
manually fit the residue in place, then use “Real-Space Refine” to get it into position.
3) As you go, use the “Mutate” or “Mutate and Auto-Fit” commands to change each Ala to
the residue you think it is, and find the correct rotamer. You can manually change
rotamers using the “Rotamers…” or “Auto Fit Rotamer” commands.
4) Once you build residues 281-326 and have assigned your best guess as to the
sequence, extract this sequence (open in PyMOL and type the command “print
cmd.get_fastastr('all')” ), then use BLAST to find the protein this is. Choose a hit that’s
761 residues long, then copy out the one-letter sequence for residues 75-326 (this
should start “FD..”).
5) Back in COOT, choose “Calculate > Mutate Residue Range…” and change the
sequence of your model (residues 75-326) to the sequence you copied above. Go ahead
and check “Autofit the mutated residues” to make your life easier. Click “Mutate”, then
give it a minute…
 6) Now go through the whole model, fix all the rotamers and run Real-Space Refine to tune
up the model to your satisfaction.
7) In the “Calculate > Other Modeling Tools…” toolbar, you can use the “Find Waters…”
command to have COOT automatically find what it thinks are ordered waters
surrounding your model. Do this only after building all the side chains you can; otherwise
it will probably build waters into side-chain density.
8) As you build, make sure you don’t cross into another asymmetric unit: In COOT, go to
Calculate > Cell & Symmetry… and turn on the Master Switch to show symmetry atoms,
and hit “Apply”.
9) COOT commands that will come in handy are shown on the last page of this document.
10) Remember to use the “File > Save Coordinates…” command early and often.
11) Return the final PDB model file to Dr. Corbett.

The Questions
1) For these electron density maps, you have a “2Fo-Fc” map colored in blue, and two
representations of a “Fo-Fc” map colored in green and blue. What are Fo and Fc, and
what do each of these maps (blue, green, and red) represent?

2) After completing your model, in COOT go to “Validate > Ramachandran Plot” and
choose your model. What does this graph show? Does your final model have any
“outliers”? Looking at the electron density, what do you see at the outlier residues (look
particularly at the main-chain carbonyl groups)?

3) In a crystallography experiment, the primary data is an electron density map. Can you
name an example of an amino acid residue whose rotamer is ambiguous in an electron
density map? How could you figure out the rotamer in this case?

4) Related to the above question, can you name a pair of amino acids that are chemically
different, yet look the same in an electron density map?

5) For the following questions, use COOT’s “Distance” tool under “Measures > Distance &
Angles” to visualize hydrogen bonds by clicking on two atoms and finding the distance
between them. Then use “Draw > Screenshot > Simple…” to output images and drop
them in below.
 First, find an alpha-helix and show the hydrogen bonds holding together one turn of the
helix (3-4 H-bonds). Paste a screenshot below

Second, show an example of H-bonding between an amino-acid side chain and a main-
chain amide or carbonyl.

Next, show an example of H-bonding between two amino acid side-chains.

Finally, show an example of H-bonding between the protein and an ordered water
molecule.

6) What type of atom (element) is present in the protein, but missing from the model? Why
do we often not bother to build this atom type?

7) Related to the above, please note the approximate resolution range at which one could
interpret the following features in an electron density map:

Alpha-helix:

Beta-sheet:

Large side-chain:

Ordered water molecule:

Hydrogen atoms:

8) Below I show a table of crystallographic “scaling” statistics for a diffraction dataset (not
this one), sorted by resolution “shells” as noted in the “RESOLUTION LIMIT” column.
For each of the following data-quality criteria, please define the criterion and then note
what resolution cutoff should be followed for this dataset based on that criterion:

RESOLUTION NUMBER OF REFLECTIONS COMPLETENESS R-FACTOR R-FACTOR COMPARED I/SIGMA R-meas CC(1/2)
LIMIT OBSERVED UNIQUE POSSIBLE OF DATA observed expected

4.50 9889 2684 2726 98.5% 2.9% 3.3% 9876 36.58 3.4% 99.8*
3.20 16774 4766 4845 98.4% 3.2% 3.4% 16769 34.09 3.7% 99.8*
2.62 22306 6158 6242 98.7% 4.0% 3.9% 22264 27.80 4.7% 99.7*
2.27 26882 7309 7391 98.9% 5.2% 4.9% 26852 22.33 6.1% 99.6*
2.03 30306 8277 8362 99.0% 7.2% 6.9% 30279 16.72 8.5% 99.3*
1.85 33978 9220 9289 99.3% 12.4% 12.8% 33963 10.05 14.6% 98.5*
1.72 37064 10070 10088 99.8% 22.8% 26.3% 37049 5.42 26.8% 94.6*
1.61 28710 10189 10797 94.4% 38.1% 45.1% 28302 2.56 47.4% 82.5*
1.51 13837 7028 11528 61.0% 59.8% 72.9% 12151 1.17 80.4% 53.0*
total 219746 65701 71268 92.2% 4.8% 5.0% 217505 13.93 5.7% 99.9*
Completeness

I/ (I over sigma)

R-factor (also called Rsym)

CC1/2

Taking all of these measures into account, what resolution would you choose as a cutoff?
Why?