Figure 3.

Flow chart of our prediction method. The method consists of seven steps and is roughly classified into two parts, [A] Appropriate Sequence Selection for Functional Region Prediction (I-VI) and [B] Cluster Detection of Conserved Residues (VII). [I] Manipulation of sequences, [II] Manipulation of a structure, [III] MSA construction of the subset of retrieved sequences and calculation of conservation scores at all alignment sites, [IV] Integration of sequence and structure information, [V] Iteration process to find the maximum DSPAC, [VI] Selection of the appropriate set of homologous sequences with the maximum DSPAC, [VII] Cluster detection of conserved residues by using LMIC.