Session 7: In silico discovery of biologically active molecules

I fought the law, but the law won

Simon Frantz, Associate Editor (News), Nature Reviews Drug Discovery

Making data on compounds available to the entire community would greatly help explorations into chemical space. However, problems such as ownership over this information create barriers to such endeavours.

With explorations into chemical space reaching a stage in which small-molecule compounds are being synthesized in greater numbers and greater diversity than ever before, there is an increasing need to select what should be synthesized and tested. To help decide this, all pharmaceutical companies, and even some academic institutions, have developed large repositories of bioactive compound analysis and screening data.

Like so many other situations, combining these resources would be much greater than the sum of its parts. Any hope that this could achieve fruition, though, seems impossible at present. The dream of having information as broadly available to the entire community as possible is tempered with the real world problems of setting up such initiatives.

For example, many at the Symposium noted that it would be difficult to maintain standards for submitted data, with Robert Pearlman of the University of Texas, and moderator of the session, saying that unintentional errors could cause problems.

But the major problem is that all data produced lies within the tight grasp of intellectual property, as companies investing the huge amounts needed to successfully discover and develop drugs require some form of protection over their compounds.

Publishing experimental testing data or predictions on any company's compound affects prior art, that is, how novel any new experimental observations are against everything that was publicly known before the invention. For this reason, it would be important for journals not to publish predictions on any company's compound, said Christopher Lipinski, formerly at Pfizer. If you find out that Compound X is active, and somebody has publicly predicted that activity, then with regards to a patent you might have a prior art 'obviousness' problem, and in the worst case scenario you might have to stop experimental work on a promising compound.

One way to create anonymity with compounds is find a way of not disclosing any predictive data that could give away the identity of the structure. But how do you mask their identities, asked Pearlman. "To say you will have some descriptors without being able to go back and identify them is like having your cake and eating it," he said.

But industry participation in such collaborative efforts is crucial, said John Schwab, a program director in the Division of Pharmacology, Physiology, and Biological Chemistry at the US National Institute of General Medical Sciences. "If we are not able to mask actual structures, then what industry participation would there be?" After the session, some chemists from industry said that they were undertaking efforts to see if they could mask compounds effectively, and although this is proving troublesome, there are initial signs that this could be achievable.

This would be encouraging news for the efforts that are currently being developed for free-access databases. Brian Shoichet, from the University of California, San Francisco, described his group's free-access database, ZINC , for virtual screening. There are usability problems, such as false negatives, but of more pressing concern is the question of how to expand the concept to other areas. Intellectual property will be the driving point behind the forward movement of these initiatives, said many delegates. One can perceive that in order for progress to be maintained, it's not just scientists that should be convening to discuss these issues, but lawyers also.

Having recognized the need for safe exchange of chemical information, Lipinski and Tudor Oprea from the University of New Mexico School of Medicine are inviting scientists and lawyers to debate this topic in March 2004 at a symposium at the 229th ACS National Meeting in San Diego.