How is MCASE different from other algorithms for quantitative structure-activity relationships (QSAR)?
MCASE stands for Multiple Computer Aided Structure Evaluation. The name gives a key to why our algorithm is unique. In all other QSAR algorithms, the user must construct a set of descriptors and then predict the activity of other chemicals based on those descriptors. The descriptors can be structural fragments or global properties (lipophilicity, molecular weight, etc.). The algorithm gives no clues about how to construct the set of descriptors, and indeed this is a difficult task. The MCASE algorithm builds a set of QSAR descriptors automatically, with no decision making on the part of the user. The only necessary inputs from the user are a list of structures along with their biological activities. Of course, the quality of the resulting database is dependent on the quality of the input data (i.e. garbage-in means garbage-out). If the input data contains a strong statistical basis for a relationship between chemical structure and biological activity, then MCASE will discover that rela
