Input: table 't' with chemical structures and molLogP column.
Output: table 't_cent' with 100 representatives with minimal
molLogP from each cluster.
make tree t full "UPGMA" split="cl" # for large tables 'full'
option can be replaced with 'compress' to avoid full distance
matrix clculation
I_out = Split( t.cluster 100 ) # 100 clusters
split t.cluster r_out
group t.cl t.molLogP "min," all "refmin" header name="t_cent"
# select centers with minimal molLogP
# group t.cl t.molLogP "max," all "refmax" header
name="t_cent" # select centers with maximum molLogP