---- User Talk ----

CLUE Cluster Analysis

T. J. O'Donnell, O'Donnell Associates

CLUE is a program to help examine the output from a clustering run. It provides a Tcl-Tk based GUI to allow the user to select individual clusters, individual compounds or groups of compounds for display in scrollable and selectable Tk list boxes, or graphically by using depict or PRADO.

Output from a clustering run is first processed by a Perl script and placed into a gdbm (GNU) data base. This data base is the source of all information for CLUE. The user can sort clusters and compounds by size, variance, molecular weight, logP, etc. Compounds can be located by name and smiles. A cross-reference list of clusters can be generated to locate all clusters in which a particular compound occurs. Groups of compounds (similar to clusters but without statistical significance) can be created and stored in the data base.

CLUE runs on a Silicon Graphics under IRIX 4 or 5 (6?). It requires Perl, gdbm and Tcl-Tk, all of which are freely available. The use of depict or PRADO is optional but highly recommended. The output from the clustering run is input into the Perl script in a format slightly modified from the Daylight Jarvis-Patrick clustering programs. Slight modification of the Perl script may be necessary for others to use CLUE.

I will demonstrate the use of CLUE.