5 minute PyMOL tutorial

I'm very lucky to have a pretty great PhD project, deep-sequencing T cell receptor repertoires. It's an interesting, exciting field, that I have only one major complaint about - I don't get to make any pretty pictures.

It sounds like a minor complaint, but it makes a difference when it comes to presentation time. My main output is sequence data; at best I produce some nice graphs, which rather lack the the impact of say, some beautiful confocal microscopy.

In order to get nice pictures to open a presentation on, I like to just fire up some molecular visualisation software and get some nice screenshots of whatever molecules are relevant to the talk. In case others wanted to do similarly - make nice pictures, without needing to become an expert in molecular visualisation or modelling - but didn't know how, here's a mini-tutorial as to how I do it.

First off, you need some visualisation software. If you're a student or teacher, I think you can get the excellent PyMOL for free, which I'll use in this blog, although there are others. I personally cut my visualisation teeth on a combination of Jmol, and (the now decidedly retro) RasMol - what you learn in one is broadly applicable to the others.

Let's get a structure to work on - head on over to the Protein Data Bank and look up something you're interested in. You can search by molecule name, ID (if you have it, you can find these in structural papers) or author name.

We need to download the PDB file for a molecule of interest, which is basically just a text file containing a set of x, y, z coordinates for all the atoms in that molecule (as described by solving the crystal structure, for example).

I've downloaded the PDB for 1MI5, a crystal structure for the LC13 TCR bound to HLAB8 complexed with a peptide from the EBNA 3A molecule of Epstein Barr Virus.

Fire up PyMOL - there's a two windows, a viewer and an interface - and open your PDB (note that if you already know the ID of your molecule, you can use this to open the file without downloading the PDB with the fetch command, e.g. 'fetch 1zgl').

At first it'll probably look like a big mess.

Orient yourself. It's pretty easy to move about with mouse; it's just hard to remember which button does what. Handily, they designers provided a nice little reminder in the bottom ride hand side.

Basically, left mouse button rotates, middle button translocates, and right mouse button zooms. Have a play, get used to it, it probably doesn't matter much anyway - if you're anything like me you'll forget it every time and have to rediscover it every time.

Now let's make it look pretty.

By default, all molecules are selected. We can use the five buttons in the top right to perform various tasks on whatever's currently selected.

The left column represents the different selections. The buttons on the right are the tasks that can be applied to those selections. Left to right: Action; Show; Hide; Label; Color

Unless I'm looking at specific interactions, I like cartoon display, as I think this makes it easiest to see generally what goes where in a structure, particularly if there's more than one polypeptide present. Click 'S' for show, then choose cartoon.

It looks somewhat fuzzy now, as it's showing the cartoon display overlain on everything else that was there. To get rid of all the fluff, go the 'H' (hide) menu, and hide the lines (default layout) and the waters (sporadic red dots).

Much better.

Now we want to be able to tell our different molecules apart. This is where the PDB web entry for our molecule really comes in handy, as it tells us what letter is used to refer to each separate molecule as. This allows us to make different selections, which we can then use to apply actions to particular molecules or subsets of molecules.

So, looking at our entry, I can enter the following code into the command line area of the interface window to generate our selections.

 select MHC, (chain a)  
 select B2m, (chain b)  
 select peptide, (chain c)  
 select alpha, (chain d)  
 select beta, (chain e)  

TCR-people, note that also for the three or four other TCR-pMHC complexes I've tried this on, they've all used the same letters for each chain, so this might be the convention.

You can then use the 'C' color button on each selection to colour them separately. I've arranged the molecule so you see down the groove of the MHC molecule, as this shows the interaction site of the whole complex well.

However, I always think it's a bit hard to see the peptide in amongst all the other molecules, so I've gone back and changed it to show as spheres.

After that, we just need to save a high-resolution image, which we can get through ray tracing:

 ray 2400, 2400  

Just save the image after ray tracing. I then cropped this down using Gimp.

All done!

When using crystal structures that you didn't produce, remember to give proper attribution; either cite the original paper, or provide the PDB ID.

For a more in-depth understanding of PyMOL, why not read the manual or check out the PyMOL wiki.