Innovative concepts for correlating molecular structures with biochemical
activities are represented by Fuzzy Logic (FL) [1-2], which mimics human
reasoning taking into account approximations and evaluations.
The FL concepts indeed provide mathematical rules and functions able to calculate
intermediate values between "absolutely true" and "absolutely false",
named degrees of membership and ranging from 0.0 to 1.0.
The main differences between classic and fuzzy logic can be, e.g.,
represented by evaluating temperatures and their membership of two
classes called Cold and Hot (Figure 1).
Figure 1. Comparison between Fuzzy and Boolean logic
by considering an example based on the concept of membership degree (µ)
of a given temperature (T) for the classes Hot and Cold.
Boolean logic is based on binary values, 0 or 1, true or false, and a
given temperature, for example 14 degrees Celsius, belongs to an only class.
Fuzzy logic substitutes continuous membership degrees for binary scores
and the classes are partially superposed. Then, the same temperature
of 14 degrees can have, for example, a membership degree of 0.3
for the class Cold and 0.8 for the Class Hot.
FL provides then interesting solutions to classification problems
within the context of imprecise categories, in which most biological
issues can be included. It represents the boundaries between neighboring
classes as something continuous, assigning to compounds a degree of
membership of each class. Recent studies [3] have also shown its ability to develop
fuzzy inference systems from numerical data, without a priori knowledge.
Then, it is able to extract relevant structure-activity relationships (SAR)
from a biochemical database.
Fuzzy Logic methods, in our laboratories, were successfully applied in the
fields of olfaction, ecotoxicity, ADME, and medicinal chemistry [4-7].
The main software developed and used by BioChemics Consulting is based on the principles of:
- Fuzzy Clustering
- Adaptive Fuzzy Partition
References
- L.A. Zadeh, Fuzzy sets and their applications to classification and clustering, in: J. Van Ryzin (Ed.), Classification and Clustering, Academic Press, New York, 1977, pp. 251-299.
- M. Sugeno, An introductory survey of fuzzy control, Inform. Sciences, (1985) 36, 59-83.
- A. Krone, P. Krause, T. Slawinski, A new rule reduction method for finding interpretable and small rule bases in high dimensional search spaces, in: Proc. of the Ninth IEEE International Conference on Fuzzy Systems, vol. 2, San Antonio, 2000, pp. 694-699.
- M. Pintore, K. Audouze, F. Ros, J. R. Chrétien, Adaptive fuzzy partition in data base mining: application to olfaction, Data Science Journal (2002) 1, 99-110.
- M. Pintore, N. Piclin, E. Benfenati, G. Gini, J. R. Chrétien, Database mining with adaptive fuzzy partition (AFP): application to the prediction of pesticide toxicity on rats, Environ. Toxicol. Chem. (2003) 22, 983-991.
- M. Pintore, H. van de Waterbeemd, N. Piclin, J. R. Chrétien, Prediction of oral bioavailability by Adaptive Fuzzy Partitioning, Eur. J. Med. Chem. (2003) 38, 427-431.
- M. Pintore, O. Taboureau, F. Ros, J.R. Chrétien, Database mining applied to central nervous systems (CNS) activity, Eur. J. Med. Chem. 36 (2001) 349-359.
|
