Dr Alan Cheng, a senior scientist at the company, told delegates at the Drug Discovery and Development of Innovative Therapeutics (DDT) conference, recently held in Boston, US, that the predictive techniques used at Amgen are then used to provide all its scientists with binding site analysis of a wide variety of proteins.

Early target analysis can help estimate how difficult it could prove to design a small molecule drug for that protein. Cheng explained that the company can then use this information to accelerate go or no-go decisions on whether a therapeutic programme is tractable. This in turn can save time and money and reduce the likelihood of failure.

He explained that there are two approaches to selectivity analysis: compound or protein based. In the first instance, compounds are grouped together based on how similar their structures are. For each group of small molecules, targets are then looked up to see if the same proteins keep cropping up.

Scientists at Amgen can then analyse the results to see if the flagged targets are either in the same therapeutic class, for example kinases involved in cell growth and death as cancer targets. In this case, the effects on other proteins might actually improve the efficacy of the drug. Alternatively, different classes of proteins may emerge from the breakdown, in which case off target effects might lead to adverse events.

If the proteins that cause these side effects can be identified, drug discovery experts can either specifically design drugs to try and avoid this cross-reactivity or use sophisticated drug delivery mechanisms to prevent the malevolent proteins from ever coming into contact with the potential drug.

The second, protein-based approach to selectivity analysis begins with a sequence comparison of the protein target with other proteins. The scientists can either measure how many amino acids in the sequence are identical or use a measure of similarity instead (based on the amino acids properties), which is clearly a more complicated approach.

Once this step is completed, the team will be left with a list of proteins that are candidates to use in a selectivity assay. Since only a small part of the protein is typically relevant to binding a small molecule, this technique can be significantly speeded up by only comparing the sequences of the binding sites. However, this obviously requires the binding site to be known and it is often more difficult to align the sequences.

If the sequences can't be aligned, the structure of the binding sites themselves can be compared instead. The scientists can calculate the minimum number of atoms that are responsible for a given proteins activity and use these pharmacophores to develop Structure-Activity Relationships (SAR). Cheng explained that this is fast enough to compare hundreds of receptors and is also useful for prioritising selectivity assays.

This is necessary because, as Cheng explained, assays return a lot of experimental data and reducing or prioritising the results that scientists need to wade through can make drug discovery significantly more efficient.

The results are often first examined by simply using a 'heat map', where protein is on one axis, drug candidate on the other and the binding affinity at each point is represented by a dot colour coded to show stronger or weaker binding.

"If we use more quantitative analysis, chemists can start to figure out which part of the drug molecule is responsible for which selectivity issue and then start to generate hybrid molecules," said Cheng.

One example of this is to generate 3D models of a drug bound to different receptors. A molecular property is measured, such as electrostatic charge, and the 3D map of this property can then be compared across different proteins "to see where both sites are 'hit' or where only one is," he added.

As computational methods become increasingly complicated - and therefore accurate, they (or their results) are also becoming more frequently used by non-specialists to inform their research, for example to prioritise their workload. This increased efficiency is hugely important in a day and age where the cost of drug development is increasing every year.