PRECISE (Predicted and Consensus Interaction Sites in
Enzymes) provides a summary of interactions between the amino acid
residues of an enzyme and its various ligands (substrate and transition state
analogues, cofactors, inhibitors, and products). In the current version this
information is extracted from the enzyme-ligand complexes in the PDB by
performing a number of steps as follows.
Clustering homologous enzyme chains. Although enzymes with the
same EC number have the same function, they may substantially differ in terms
of sequence and/or structure. Since consensus binding sites can be defined only
for enzymes with appropriate overlap of their sequences and structures, we have
clustered the enzymes in the PDB such that in each cluster the proteins have
the same EC number, and all chains are sequence-similar, with a BLAST p-value
of 10e-40. The clusters are based on chains rather than the entire protein,
since enzymes may have chains of different functions and/or non-homologous
Selecting a representative for each homologous cluster. The
chains within a sequence-similar cluster thus derived are automatically ranked
according to the precision and completeness of their structural data. The
measures of the structural quality are adopted from the NCBIâ€™s Non-Redundant
PDB Chain Set (nrpdb), see
Selecting ligand type. If the PDB file contains a ligand, it
is classified as (a) peptide, (b) nucleotide, (c) metal ion, or (d) â€œotherâ€?.
Calculation of hydrogen bonds and non-bonded interactions. The
receptor-ligand interactions are determined using the program HBPLUS by
Thornton and coworkers (http://www.biochem.ucl.ac.uk/bsm/hbplus/home.html).
The results are later parsed and extracted into residue-based formats, giving
separate files for each residue that is interacting with the ligand.
Summing all interactions within a cluster for the aligned residues.
The number of â€œhitsâ€?, or interactions, for a given residue is calculated
automatically throughout the entire cluster. Again, the hits are atom based;
two interacting residue can have a hit greater than one if several atoms are
involved in the interaction.
Final output. If there are non-homologues chains present in
the query pdb, a subsequent page will let users to specify which chain to show.
The output page will show the sequence of the representative of the cluster
along with different color codes for each residue representing the number of
hits. Optional filtersallow users to restrict the output to (a) selected chains
in the cluster (b) non-bonded or hydrogen bonding interactions; and (c)
selected ligand types.