Hit picking party

From DISI
Jump to: navigation, search

Before we buy compounds, we have a hit picking party.

Before the party

The investigator who performed the virtual screen looks critically at the results, selecting perhaps a dozen or two interesting compounds from among the top 500. Several copies of the top 500 list are printed, sometimes with additional supporting documentation, and distributed to the participants. Check CSD data for compound pose information. The pKa of each molecule can be checked. Think about reactive groups for molecules. Think about PAINS http://pasilla.health.unm.edu/tomcat/biocomp/smartsfilter

During the party

We take a good look at the site, and any crystallographic ligands or known experimental ligands. Think about enthalpic and entropic contributions to the binding. Think about possible water structure. Think of receptor flexibility, especially His, Gln, Asn, Ser, Tyr, Thr. Look for charged residues. Look at the surface colored by charge to look for greasy patches.

Look at one ligand at a time, and rate each compound (e.g. 1, 2, or 3 checkmarks). People speak out, making favorable and unfavorable comments about each ligand. Broken molecules are rapidly ignored.

Watch out for very high or very low calculated LogP. Watch for weird conformations and weird protonation states and weird tautomeric forms. ZINC is imperfect, use your judgement. Watch out for super floppy molecules. Watch out for buried polarity or missed opportunities. hydroxide is the worst. If hydroxide ain't happy ain't nobody happy.

After the party

The investigator purchases compounds as discussed and has them tested experimentally. Ideally, a report of the fate of each compound can be shared with other members of the group within a few weeks. We try to test at as high concentration as the assay will allow. We like to order 10mg so that we can test for purity, identity, and also repeat the experiment.

Epilogue

Sometimes we go through several iterations. If no hits arise, we go back to the drawing board. We try to incorporate all available information, both in the modeling, and in the judgement of the ligands. Watch out for aggregators, and be sure to control for these during the experiments. [1]