Table 2. JET2 automated clustering algorithm.
First round to detect main clusters | |
if #(T JET > 0.6)/N < 0.3 * f intfrac(N)—————– | • Test whether the evolutionary signal is too low: N, number of surface residues f intfrac(N), expected size of the interface #(T JET > 0.6)/N, proportion of highly conserved residues |
then | |
choose SC3; | |
else | |
choose SC1; | |
end | |
detect cluster seeds; | |
if SC1 then | |
if Seeds = ∅———————————————— | • Test whether SC1 found some seed: Seeds, set of seeds |
then | |
choose SC3; | |
detect cluster seeds; | |
else | |
if for all s ∈ Seeds—————– | • Test whether the seed conservation signal is homogeneous: σ(s) and μ(s), dispersion and mean of the distribution of the scores computed over the residues in seed s σ surf and μ surf, dispersion and mean of the distribution of the scores computed over all surface residues |
then | |
choose SC2; | |
detect cluster seeds; | |
end | |
end | |
end | |
add the extensions to the seeds; | |
add the outer layers to the clusters; | |
if SC1 or SC2 then | Second round to detect additional clusters using a complementary SC |
choose SC3; | |
else | |
choose SC2; | |
end | |
detect cluster seeds; | |
add the extension to the seeds; | |
add the outer layers to the clusters; | |
combine clusters; | Merge main clusters with additional clusters <5Å away |
scoreMax ← maxr ∈ C(score(r)); | Extension of a cluster C |
while do | |
newRes ← {}; | |
for r ∈ {neighbors of C} do |
score(r), score of the residue r
μ(C), mean score computed over cluster C , threshold score for residues , threshold score for clusters {neighbors of C}, ensemble of residues <5Å away from C |
if then | |
add r to newRes; | |
end | |
end | |
add newRes to C; | |
scoreMax ← maxr ∈ newRes(score(r)); | |
end | |
newRes ← {}; | Addition of an outer layer to a cluster C |
for r ∈ {neighbors of C} do | |
if then | {neighbors of C}, ensemble of residues <5Å away from C |
add r to newRes; | |
end | |
end | |
add newRes to C; |