A new screening process based on tree-structured data has been constructed for an in-house chemical substance
database. The data structure and algorithm have improved the selectivity of screening and reduced search
time. Thus, a prescreening step selecting an appropriate key is introduced to accelerate the total screening
process. Its effectiveness has been proved by the comparison with the search without such prescreening.
The present tree-structured system has been compared with a fragment-oriented screening system along
with the retrieval of 87 query structures. The result shows excellent performance of the present screening
system; it is because the tree-structured data include structural information of all nodes of the structures.