tanimoto

The similarity search uses a tanimoto coefficient in order to calculate similarity of two structures (respectively their fingerprints). This is realized for mysql by use of a user defined function (udf), which is written in C and must be installed in order to run similarity searches. This is the only bit which is probably not easily ported to other rdbms. Follow these step to install the udf.

• You need the mysql header files. In debian, they come as part of a package called "libmysqlclient-dev"

• Change to the src/c++ directory and perform "g++ -shared -fPIC -o tanimoto.so tanimoto.cc".

• Copy the file tanimoto.so to the directory for shared libraries, usually /usr/lib on Linux systems.

• Call create function tanimoto_coefficient RETURNS REAL SONAME "tanimoto.so"; in mysql

• If you do a select * from mysql.func; the function should show up.