Reviews in Computational Chemistry Volume 18
Reviews in Computational Chemistry Volume 18
Reviews in Computational Chemistry Volume 18
- No tags were found...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
64 The Use of Scor<strong>in</strong>g Functions <strong>in</strong> Drug Discovery Applications<br />
biological test<strong>in</strong>g. This subset typically consists of ca. 100–2000 compounds<br />
selected from libraries conta<strong>in</strong><strong>in</strong>g 100,000–500,000 compounds. Therefore,<br />
it is essential that the computational process <strong>in</strong>clud<strong>in</strong>g the scor<strong>in</strong>g function<br />
is fast enough to handle several thousand compounds <strong>in</strong> a short period of<br />
time. Consequently, only the fastest scor<strong>in</strong>g functions are currently used for<br />
this purpose. Speed is especially important for those scor<strong>in</strong>g functions used<br />
as objective functions dur<strong>in</strong>g the dock<strong>in</strong>g calculations, s<strong>in</strong>ce they are evaluated<br />
several hundred to a thousand or so times dur<strong>in</strong>g the dock<strong>in</strong>g process of a<br />
s<strong>in</strong>gle compound. <strong>18</strong><br />
Follow<strong>in</strong>g a successful virtual screen<strong>in</strong>g run, the selected subset of compounds<br />
conta<strong>in</strong>s a significantly enhanced number of active compounds as compared<br />
to a random selection. A key parameter to measure the performance of<br />
dock<strong>in</strong>g and scor<strong>in</strong>g methods is the so-called ‘‘enrichment factor.’’ It is simply<br />
the ratio of active compounds <strong>in</strong> the subset selected by dock<strong>in</strong>g divided by the<br />
number of active compounds <strong>in</strong> a randomly chosen subset of equal size. In<br />
practice, enrichment factors are far from the ideal case, where all active compounds<br />
are placed on the top ranks of a prioritized list. Insufficiencies of current<br />
scor<strong>in</strong>g functions, as discussed <strong>in</strong> the previous section, are partly<br />
responsible for moderate enrichment rates. Another major reason is the fact<br />
that the receptor is still treated as a rigid object <strong>in</strong> the computational protocols<br />
be<strong>in</strong>g used. To generate correct b<strong>in</strong>d<strong>in</strong>g modes of different molecules, it is<br />
necessary to predict <strong>in</strong>duced fit phenomena. Unfortunately, predict<strong>in</strong>g prote<strong>in</strong><br />
flexibility rema<strong>in</strong>s extremely difficult and computationally expensive. <strong>18</strong>9–196<br />
Seed<strong>in</strong>g Experiments<br />
Enrichment factors can be calculated only when experimental data are<br />
available for the full library. But only a few libraries conta<strong>in</strong><strong>in</strong>g experimental<br />
data that have been measured under uniform conditions for all members are<br />
available to the public. Several authors have therefore tested the predictive<br />
ability of dock<strong>in</strong>g and scor<strong>in</strong>g tools by compil<strong>in</strong>g an arbitrarily selected set<br />
of diverse, drug-like compounds and then add<strong>in</strong>g to it a number of known<br />
active compounds. This ‘‘seeded’’ library is then subjected to the virtual<br />
screen, and, for the purpose of evaluation, it is assumed that the added active<br />
compounds are the only true actives <strong>in</strong> the library. Several such experiments<br />
have been published. An example is a study performed at Merck with the<br />
dock<strong>in</strong>g program FLOG. 59 A library consist<strong>in</strong>g of 10,000 compounds <strong>in</strong>clud<strong>in</strong>g<br />
<strong>in</strong>hibitors of various types of proteases and HIV protease was docked <strong>in</strong>to<br />
the active site of HIV protease. This resulted <strong>in</strong> excellent enrichment of the<br />
HIV protease <strong>in</strong>hibitors: all <strong>in</strong>hibitors but one were among the top 500 library<br />
members. However, <strong>in</strong>hibitors of other proteases were also considerably<br />
enriched. 197<br />
Seed<strong>in</strong>g experiments allow for comparisons of different scor<strong>in</strong>g functions<br />
with respect to their performance for different targets. Seed<strong>in</strong>g experiments