ABSTRACT
CASP: critical assessment of methods for protein structure prediction; an international experiment to evaluate the accuracy of methods for structure prediction; methods are tested using a set of proteins before their structure is publicly available
Core: part of the structure that is structurally conserved between two homologous proteins
Random coil: configuration of a protein in the unfolded state
r.m.s.d.: root mean square deviation; defined as: , where xi and xi′
are the coordinates of the corresponding atoms and N is the total number of superimposed atoms
Test set: set of data not used in the training set, and left aside to assess the reliability of a method
Training set: set of data used to parameterize (or train) a program-for example, an artificial neural network
We have mentioned several times that one of the main interests of bioinformatics is to assign a function, possibly at the molecular level, to gene products. In other words, we want to associate each gene to its own product and unravel the molecular mechanism responsible for the function of the latter. Ideally, if we are dealing with enzymes, we would like to identify the groups involved in catalysis as well as those involved in substrate recognition.