For molecular identi®cation of essential catalytic residues in the
active site of the cassava linamarase, site-directed mutagenesis
was carried out [29]. As a search for possible structural changes
introduced into the active site by the above amino acid substitutions,
molecular modelling was carried out, using the MODELLER
program [30], for the cassava linamarase on the basis of
the white clover linamarase structure [13] utilizing the 49%
amino acid sequence identity and 61% homology between the
two proteins. The model generated had a root mean square of
differences of 0.085 A / for the superimposition of Ca atoms and
therefore satis®ed the criteria of a structure at 2 A / resolution.
The model was used subsequently to build the mutated active
sites using conformational search and potential-energy
minimization (Figure 3).