Computer Chemistry Consultancy - XML4Pharma, Katzelbachweg 18, A-8052 Thal, Austria, info@CompChemCons.com

Intranet Implementation of Neural Networks

Many chemical and pharmaceutical companies have developed neural networks in the last 10 years. However, in many cases the developed neural networks have not been properly deployed over the people that should have access to this knowledge. In many cases the neural networks have been implemented in specical computer programs that are available to, and can only be handled by a few specialists. So those that SHOULD primarily have access to the developed knowledge, cannot get it in an easy way.

Computer Chemistry Consultancy is a pioneer in the intranet deployment of neural networks. For a number of chemical companies and their business units, a number of neural networks implemented on their intranet:

  • The XLogP program for the calculation of hydrophobicity - LogP(octanol/water) developed at the Molecular Design Laboratory of the Peking University, which at that time was only available for Silicon Graphics (Unix) computers, was interfaced to the intranet so that every researcher in the pharmaceutical company could perform a LogP calculation through the intranet. Before, researchers had to send a request (written or by mail) to the "owner" of the SGI machine and hope they would soon get an answer.

  • For a surfactants-developing company, a neural network was deployed on the intranet for the prediction of biodegradability of new surfactants. This greatly reduced the cost of development, as before the neural network was developed, it happened many times that weeks or months were spend on synthesizing the new material, and than it was found that it was too less biodegradable to be commercialized. The intranet tool was username-password protected, so that only people who had permission, had access to the information.

  • For another company, an intranet tool was developed to calculate the Critical Micelle Concentration (CMC) of a class of surfactants, using a QSPR (Quantitative Structure Property Relation). Although the QSPR was described in public literature, the company did not have the software-development skills in-house to develop an intranet application, so we did it for them.

Notes:

  • Many people believe that the intranet is only suitable for "company-public" information. This is however not true at all! Information on the intranet can very easily be secured so that only people with permission to view that information have access to it.

Computer Chemistry Consultancy, June 2001