The ligand-similarity search with Screen3d consists of the following main steps (scripts referred to can be copied from /sc/hydra/projects/mezeim01b/CHEMAXON_util):
Pre-processing a library
For a library LIB, the pre-processing generates three files: LIB.bin, LIB-name.bin and LIB-storage.bin.
To preprocess a library file LIB, run the script prep_screen3d_job.csh:
If the ligands are gathered in a set of ligand aggregate files, prep_screen3d_job.csh has to be run on each
Running the similarity search
To run a search on a (pre-processed) library LIB, molecules that are similar to LIG (present in the work directory as LIG.mrv) run the script screen_job.csh as follows
bsub -L /bin/csh -J LIB -R "rusage[mem=16000]" -o LIB.LIG.out -e $LIB.LIG.err screen_job.csh LIB LIG /sc/orga/projects/liglib/screen3d/bin"
The ZINC library has been broken up into 178 separate files. To run a screen against ZINC, run the script screen.csh and answer the quizzes - it will submit as many jobs as there are files in the directory ZINC.
The screening jobs are not very robust and there is a good chance that some fo the jobs fail. The script clean.csh detects and removes the failed jobs; running the screen.csh script again will only submit those jobs that have failed. There is also a script checklog.csh that monitors the status of the screening jobs.
Post processing
The screening of a ligand library LIB against a ligand LIG results in a file LIB-LIG.log. The file ends with the Tanimoto scores of the molecules in the library, sorted in decreasing score.
When the library consists of several files, the script anal.csh can combine the Tanimoto list and sort them. For screening the ZINC library, there is also an option to extract the .sdf files of the top-scoring ligands.
All scripts referred to are in the directory /sc/hydra/projects/mezeim01b/CHEMAXON_util