The greatest wish of many of us suffering from a common cold, as well as those with more serious diseases, could be summarised in the 80’s pop hit by Huey Lewis and The News: “…I want a new drug, one that won’t make me sick, one that won’t make me crash my car, or make me feel three feet thick…” Finding new drugs is an extremely difficult and expensive task though.

A recent study estimated that around half a billion to two billion dollars of investment is needed to bring a new drug to market. The pharmaceutical giant Pfizer, for example, spent over one billion dollars researching and developing its blockbuster cholesterol-lowering drug Lipitor. In addition, a promising candidate drug can be discarded even once it gets as far as clinical tests on human volunteers, at huge cost for the pharmaceutical company developing it. Pharmaceutical firms are therefore attempting to reduce the risk of bankruptcy by making use of computer-assisted theoretical models for drug development instead of wasting resources and money in the real world.

Before describing these approaches, it is important to understand how drugs work. Strictly speaking, a drug is a small molecule that interferes with the function of other molecules—proteins in most cases—by interacting with them and modulating their deleterious activity, thus restoring our health. Proteins behave as a ‘lock’—their specific three-dimensional structure means they can only interact with substances that fit precisely, the ‘keys’. Computer-assisted modelling approaches can be useful in finding out the three-dimensional structure of the lock and then in finding the right key in a haystack of potential drugs.

Experimentally determining the three-dimensional structure of a protein can be a painful and time-consuming task. The good news is that computers can help us. The three-dimensional structure of a protein is determined by the sequence of its building blocks, amino acids, which are bound together like pearls in a necklace. Each specific amino acid sequence produces a unique three-dimensional protein structure. Unlike characterising the three-dimensional structure, determining the amino acid sequence of a protein is a trivial task.

Imagine we have the amino acid sequence of a promising protein target and we want to predict its structure. If we know the structure of a protein that has a similar amino acid sequence, we can use this information to deduce the structure of the target protein. Nowadays we do this by feeding the amino acid sequence into a computer programme. The programme compares the amino acid sequence to the sequences of proteins that we already know the structure to and, after a bit of number crunching, the shape of our target protein is returned.

In my lab at the Department of Biochemical Sciences in the Sapienza University of Rome, we have applied such techniques to predict the structure of several proteins that could be important for the development of therapeutic agents, including drugs that could be used to treat tumours, infections, malaria and neurodegenerative diseases.

Once we’ve found the lock, how do we find the keys? The ‘keys’ are stored in huge databases of drug candidate molecules called virtual libraries. Computational techniques called ‘virtual screening approaches’ can be used to find among these compounds the one(s) that fit precisely in the ‘lock’—the protein of interest.

Experimentally screening every single molecule in the library, one by one, could easily take a whole lifetime. Luckily, computers can again do the dirty work for us and sift through all the potential ‘keys’. If everything goes well, we end up with a ‘hit compound’: a small molecule able to bind to and interfere with a therapeutically relevant protein target. The way from here to the development of a new drug is full of pitfalls and drawbacks, however, and too often ends with the sad refrain of the hit song by The Verve: “The Drugs Don’t Work.” In fact, one of my postgraduate students, who’s currently testing some potential treatments for Parkinson’s disease that were identified by computational modelling, sometimes lounges around humming this theme...

There are some success stories in drug discovery though. A stunning example of success (and serendipity…) is the tale of one of the most successful drugs in history, Viagra. This drug was designed to relieve the symptoms of angina and to cure high blood pressure. Initial clinical trials suggested that this drug had little effect on angina, but that it could induce an intriguing side effect: penile erection! At least for some lucky elderly gentlemen, the wish came true!

So if you’re stuck in bed with a terrible cold and only your laptop for company, don’t forget that someone out there is being much more proactive and making use of computers to design drugs that will make you stop blowing your nose.