CHEM 451: Data & Analysis

Overheard in the library...

Biochemistry student: "Ouch, I've been hacking at this terminal for hours. My head really hurts."

Accounting student: "Take an analgesic; you'll feel better. It works for me."

The OTC analgesics market in the United States is expected to generate a revenue of $8.07 billion dollars in 2025 (analgesics market estimate). That's a per capita annual spending of $23.48. While this isn't a business course, it is probably safe to say that if you could tap into a market where every man, woman and child in the United States spends over twenty dollars every year on your products it would be a good business. Hey, who are we kidding, if you could get just a fraction of that steady business life would be good.

To know where you're going, it helps to know where you've been. The bark and leaves of various plants have been used to treat chronic, intermittant pain for thousands of years. Most of those natural products contain salicylates (benzoic acid with an ortho hydroxyl group). Felix Hoffman, working at Bayer, synthesized acetylated salicylic acid in 1897. The product was named Aspirin and went on to become the most widely used medicine, probably of all time, in the western medical tradition.

Acetaminophen was first synthesized in 1878 and used to treat pain in 1893. However, it wasn't used widely until the introduction of the prescription medicine Tylenol Elixir for Children in the 1950 as a non-aspirin analgesic option. Ibuprofen was developed in the early 1960s at Boots Pure Drug, Nottingham, UK. It's said that Stewart Adams, the leading pharmacologist in the development of ibuprofen, first used the molecule to relieve his discomfort induced by a hangover—a very English problem.

Aspirin, acetaminophen and ibuprofen target, among other enzymes, cyclooxygenase–1 (COX–1, also known as prostaglandin endoperoxidase H synthase. The function is similar toward the related enzyme COX–2).

Aspirin is a non–reversible inhibitor of COX–1 (no competitive, uncompetitive or mixed for aspirin). Particularly, aspirin acetylates serine 530 in human COX–1, rendering the enzyme unable to catalyze any more turnovers. This mechanism of inhibition is known generally as a suicide inhibitor, the drug is a substrate which the enzyme begins the catalytic mechanism with, but somewhere in the middle of the reaction, surprise, the enzyme is (usually) covalently modified by the inhibitor. The drug blows itself up to destroy the enzyme.

Recognizing that this fertile market might have room for a new competitor, wouldn't it be good if you could find a novel, reversible inhibitor of COX–1?

That's where a large(–ish) scale docking experiment might help to find a set of candidates for further study.

To begin, create a new directory (trust me, you'll be making a large number of files, you'll want to keep them isolated in your file system). Download all of these common files to that directory.

6y3ca.tar.gz contains the structure of the A chain of COX–1 from the structure 6Y3C and the parameters file which defines the docking site to search on the receptor along with software configuration.
pscrp (Preform SCReen Parallel) is the master control program that will run the screen over every potential ligand for you. Yes, it is (almost) that easy.
pro-scr (PROgress of SCReen) is program that will monitor the progess of the screen through ligands. This one is optional; use it if you're curious about your progress. Otherwise wait.
rscr (Report SCReen) is a program that will mine the results and link potential ligands to the predicted binding energy.

Untar and/or unzip each of these files. That's something like:

As a warning, you may have to set the execute bit on the programs in order to allow them to be executed. chmod +x is your friend here.

Now comes the unique part. Prepared for each of you is an archive with 10,000 small molecules (all less that 300 g/mol) selected at random from the ZINC 20 database. Download the archive which matches your student id number and unpack.

When you're finished, you will have five files in your directory along with 10,000 directories. This is good. In order to start the screen, and keep it running as you logout, try this syntax:

There is lot going on in this command line; a little explanation is in order. nohup (short for no hang up) is a program which starts your job such that it is not owned by your shell. This allows it to continue running after your shell exits. The control program reads the current directory (and subdirectories) looking for ligands and writes the input syntax to the docking program (try ./pscr at the prompt so you can see the commands to be issued as harmless text). The screen then uses xargs to control the jobs (four running in parallel, -P 4, in this example). The docking will be done by vina which is a relative of AutoDock but is much faster in docking. The last bit redirects output down the rabbit hole of the null device (> /dev/null) so you don't see it, and places your job in the background (&) so that your prompt is available to use while this (long) job runs.

After a long time (you can use ./pro-scr from the experiment directory to check progress; each molecule should take about a minute to dock, that's 60 per hour or 1440 per day or just about 7 days on one processor), you can summarize your results with a command like this:

Where the first column is the ligand and the second column is the calculated binding energy in kcal/mol, lower numbers are better.

Which five molecules have the lowest calculated binding energies (note: the docking screen generates ten model poses for each potential ligand, you may see the same ligand more than once in the top five of your output, please select five unique molecules)? Prepare a table of those molecules including structure (search for the molecule using the directory number (ignore the -1 or similar parts) at ZINC 20 Substances, use the top input box) and best predicted ΔG? Do your molecules appear to have structures or substructures in common?

Prepare a figure showing COX–1 and your best posed ligand (note: in the directory for each ligand there is a file with the ligand number and the *.dk extension. This file holds the energies; model 1 is the lowest energy. In the file out.pdbqt, you'll find coordinates for each mode. Be sure to use model 1 in your MolScript input). Please make it a plot that clearly shows the interaction of the ligand and receptor.

How many hydrogen bonds are formed between the receptor and the ligand? How much surface area is buried by the ligand?