International Journal of Chemoinformatics and Chemical Engineering, 2(2), 1-14, July-December 2012 1
Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
Keywords: Automating Classiication, Bond-Electron Matrix, Chemical Reaction Classiication, Reaction
Matrix, Ugi’s Scheme
INTRODUCTION
One of the major tasks of chemoinformatics is
to predict the outcome of a chemical reaction.
For this purpose the first step is to classify exist-
ing chemical reactions, on the basis of which
unknown reactions can be predicted. In the mid
1970s, Ugi and his co-workers developed a tech-
nique to represent chemical reactions by means
of Reaction matrices (R-matrices) (Dugundji &
Ugi, 1973; Ivanciuc, 2010). Later researchers
had studied (Bart & Garagnani, 1977) several
reaction schemes and finally, Ugi’s work, which
classifies chemical reactions into 30 different
classes, came to be known as Ugi’s scheme
(Bart & Garagnani, 1977; Gasteiger & Engel,
2003). In this paper we have proposed an ef-
ficient algorithm to automate classification of
chemical reactions based on Ugi’s scheme.
In the following section we first describe the
background to elaborate the representation of
molecules and chemical reactions using graph
theoretic techniques - Bond-Electron matrix
(BE-matrix) and Reaction matrix (R-matrix)
An Eficient Algorithm for
Automating Classiication of
Chemical Reactions into Classes
in Ugi’s Reaction Scheme
Sanjay Ram, Bengal Engineering and Science University, Shibpur, India
Somnath Pal, Bengal Engineering and Science University, Shibpur, India
ABSTRACT
There are two approaches for classiication of chemical reactions: Model-Driven and Data-Driven. In this
paper, the authors develop an eficient algorithm based on a model-driven approach developed by Ugi and
co-workers for classiication of chemical reactions. The authors’algorithm takes reaction matrix of a chemical
reaction as input and generates its appropriate class as output. Reaction matrices being symmetric, matrix
implementation of Ugi’s scheme using upper/lower tri-angular matrix is of O(n
2
) in terms of space complexity.
Time complexity of similar matrix implementation is O(n
4
), both in worst case as well as in average case. The
proposed algorithm uses two ixed size look-up tables in a novel way and requires constant space complexity.
Time complexity both in worst and average cases of the algorithm is linear.
DOI: 10.4018/ijcce.2012070101