Computer Aided Drug Design


Alex Habeger
    Designing drugs can be a meticulous and time consuming process. Time is money in drug development and computers can help reduce costs greatly. One of the earliest examples of computer aided drug design (CADD) is from 1985. A researcher took a part of the influenza virus and after computerized examination he found where influenza breaks through cell walls. The program that discovered that location is GRID. GRID works by examining all the different surfaces of a molocule for locations where a strong bond is possible. A drug was soon designed and synthesized that would bond well to this location. The part of the influenza virus was a neuraminidase protien and the drug developed was Tamiflu.

Virtual High-Throughput Screening

    Conventional drug discovery methods are often by guess and check. Virtual High-Throughput Screening (vHTS) is a computerized method of guess and check. In the process of vHTS a library of hopeful compounds is supplied to a computing resource where each compound is examined on certain criteria. The compounds are ranked based on their simulated effectiveness. The results are returned much faster than real world testing due to the lack of procedure involved. Millions of compounds can be tested in a day. In 2001 vHTS was benchmarked at "one minute of CPU time per ligand per processor". This is much more than standard testing would allow, increasing the odds that a better candidate is found and also decreasing the need for costly testing of unsuccessful compounds. The simulated environments are perfectly clean and the synthesis of every compound is instant. Computer simulations are not only useful for finding new drugs that work on old conditions, it can be used to find why bacteria are now resistant to old drugs.
    There are problems with vHTS. The best drugs that vHTS can find are only as good as the compounds in the database that is supplied to it. There is no one singledemo of how drugs interact with receptors database of all known compounds, they are often bought from chemical suppliers. There is work being done to take vHTS to the next level. One area where major work is being done is designing logic to first examine the compounds that vHTS rates the best. Then the logic would find why they work as well as they do. This delves deeply into the realm of structure based drug design and genetic algorithms. Some of the compounds in the database can also be not realistic. Current logic only rates compounds on their effectiveness. These compounds can be impossible or expensive to synthesize. They can also have undesired side effects or simply be insoluble and unable to be delivered.
    vHTS involves many relatively short computations that are well suited to cluster environments. Projects have been done by several computers in a private apartment, an internet wide project, to more standard environments. Each simulation can be ran independently of all others with minimal data needed from a central  shared source. This lends to the idea that chemical simulations are best ran on massively parallel clusters with a star network topology. Higher order networks do little to improve performance, lending even more support to commodity hardware clusters, or even batch processing.

People in the field

    Being a in the middle ground between different disciplines computer aided drug design has difficulties finding people with an appropriate skill set. Most come from computational chemistry backgrounds. The remainder come from conventional or "wet" chemistry, organic, or medicinal backgrounds. Experience in conventional chemistry can be quite valuable as it gives a greater insight into translating the virtual world into reality.

How compounds are ranked:
http://www.idbs.com/PredictionBase/PredictionBaseHTSFilter.asp
2 million compounds to 100 using vHTS:
http://focus.hms.harvard.edu/2005/Mar11_2005/drug_discovery.shtml
http://www.dddmag.com/...
3.57 billion compounds to 300,000 using vHTS
http://www.chem.ox.ac.uk/anthrax/
Different types of CADD:
http://www.b-eye-network.com/view/852
Commercial products catering to CADD:
http://www-03.ibm.com/press/us/en/pressrelease/6427.wss
http://www.prnewswire.com/...
http://www.albmolecular.com/discovery/cadd.html
http://www.csiro.au/csiro/content/standard/psi6,,.html
http://www.computers.us.fujitsu.com/www/solutions.shtml?solutions/life_sciences
http://www.cato.com/biotech/bio-genomics.html
List of software catering to CADD/CAMM:
http://www.netsci.org/Resources/Software/Modeling/CADD/
Computer aided drug testing:
http://pubs.acs.org/subscribe/journals/mdd/v06/i02/pdf/203willis.pdf
Lecture on structure based drug design:
http://www.chemie.uni-erlangen.de/clark/multimedialabor/_lectures/drug_design_3.html
IBM benchmarks vHTS process and results:
http://www.research.ibm.com/journal/sj/402/waszkowycz.html