Chemistry Reference
In-Depth Information
Automated techniques are necessary to manage and analyze the huge amount of
data in chemical/biological databases. KBDD (Knowledge-based drug discovery)
can be used to extract from databases useful and novel knowledge units. The core
of the KBDD process is data mining that involves applying algorithms to explore
data and discover significant patterns yielding 'knowledge units' [324-338].
Effectively, knowledge discovery (data mining), is a computer-assisted process
which consists of digging through/analyzing huge sets of data. Subsequently, the
'meaning' of the data is extracted. Effectively, these tools can predict behaviors as
well as future trends. This can help scientists to make knowledge-driven
decisions. Data mining tools can yield many answers that are too time consuming
traditionally. They go through databases for hidden patterns. It is possible to find
predictive information (outside expectations) that experts may miss. The name has
its origin from the similarities between searching a large database for valuable
information and mining for a valuable ore in a mountain. Both processes require
sifting or intelligently probing through considerable material to find answers.
KBDD, for example, can enhance 3DVS by applying machine-learning
techniques to 3D information regarding ligands, proteins and their corresponding
interactions. Nonetheless, the KBDD process needs training on a data set (active
and inactive compounds). The automation of this training set is not evident.
KBDD can be used with protein-ligand interactions for the following purposes: a)
design scoring functions, b) target focused libraries, c) target databases of known
active ligands to derive physicochemical, structural and ADME-Tox property
profiles [324-338].
As an example, absorption, distribution, metabolism, excretion, and toxicity data
is important for developing reliable
in silico
ADMET models. Contributing to this
effort a pharmacokinetics knowledge base (PKKB) database was reported which
compile comprehensive information about ADMET properties into a single
electronic repository. This database contains more than 10
000 experimental
ADMET measurements of 1685 drugs, including octanol/water partition
coefficient, solubility, dissociation constant, intestinal absorption, Caco-2
permeability, human bioavailability, plasma protein binding, blood-plasma
