D-EXP010

=Objective= Rank a virtual library of Ugi products using Rajarshi Guha's algorithm predicting anti-tumor activity.

=Procedure= UClib01 was processed using the Guha [|algorithm] offline.

=Results= A zip file of the results is available here (zip file) txt file (file currently not available but can be reproduced if needed)

The file is tab delimited, the first column is the SMILES for which the prediction has been made. The next 60 columns represent the prediction for each of the cell lines (whose names are given in the headers). The numbers are the probability of being active (i.e., showing anti-cancer activity) for a given cell line. The number ranges from 0 to 1, 1 indicating that it is active against the cell line and 0 indicating that it is inactive. The closer to 1 the more likely it is active. A value of 0.5 would indicate that the model is not sure whether it is active or inactive. One way to prioritize molecules that have > 0.7 probability of being active in at least 1 of the cell lines. This gives 21 molecules, whose [|SMILES are here (RajPriority01)]. (note that ChemSketch can only open the first 13 molecules but all are valid apparently)

=Discussion= All of the top 21 molecules use naphthyl isocyanide and the first 13 use boc-methionine. Here is the first of the RajPriority01 file as an example:

=Conclusion= All of top 21 molecules used naphthyl isocyanide, which was not available for immediate shipping. Another run will be done with a larger library only containing starting materials that are immediately obtainable.