r one of the C terminal residues. The cleavage happens between

ଵ

ᇱ. The requirement of a successful interaction between a protease

avage site in a polyprotein is the amino acid distribution trend

ing a cleavage site. Different proteases require different patterns

mino acid composition trend of a cleavage site. For instance, the

protease cleavage usually happens when ܴଵൌܴ, where R is the

id arginine [Yang, et al., 2006]. The trypsin protease cleavage

happens when ܴଵൌܭ or ܴଵൌܴ, where K is the amino acid

homson, et al., 2003].

uccess of the protease cleavage pattern discovery for a special

therefore depends on the accurate recognition of the amino acid

ion trend within a set of cleaved peptides. Only when this

d pattern has a great discrimination power between cleaved

and non-cleaved peptides, which is demonstrated in a machine

model, following successful inhibitor design against the protease is

ble [Goodman, et al., 1983; Sheth, et al., 1984; Singh, et al., 2001].

inear discriminant analysis algorithm

alysing a biological data set, it is often required to recognise the

of molecules through analysing the factors associated with the

of molecules, such as predicting protein function, gene-gene

n, enzyme cleavage sites and functional genes. This process is

e object classification. When there are only two classes of objects

set, this process is called a discriminant analysis. The final

of a classification model is to label novel data (novel objects) or

ptides in this context.

inear discriminant analysis algorithm (LDA) was developed by

isher as a linear model for classifying data and is also called the

scriminant algorithm [Fisher, 1936]. The earliest application of

PubMed was a perinatal mortality study [Greenberg, 1963].

LDA is a linear model, it lays down the foundation of

tion analysis and it still attracts a significant attention in pattern