Figure 1.
Graphical overview of the ASC method.
(A) In the first step training sequences are aligned using 3DCoffee to get an MSA. (B) In a second step residues lining the active site are extracted from the template structure. (C) The third step maps the extracted residues along the MSA to get a signature of the active site for each sequence. (D) These signatures are then encoded into feature vectors using the three descriptors . Alternatively, kernels may be used. (E) The final ASC model is trained using the generated feature vectors.
Table 1.
ASC performance on Hannenhalli benchmark.
Table 2.
ASC performance on Capra benchmark.
Table 3.
Decarboxylating dehydrogenases active site signature.
Figure 2.
View on the superimposed active sites of IPMDH and ICDH.
The first chain of the homo-dimeric enzyme is represented by its solvent-excluded surface. The second chain is depicted in a backbone representation. The two substrates isocitrate (purple) and isopropylmalate (green) lie in the interface of the two chains. IPMDH sidechains are coloured green and sidechains from ICDH (PDB-Id: 1AI2, [40]) are coloured purple. This figure was created using BALLView [41].