[Zinc-fans] missing NCI compounds in zinc?

John Irwin jji at cgl.ucsf.edu
Mon Aug 22 10:30:38 PDT 2005

Hi Jose

Thanks for your email and your interest in ZINC. Despite appearances to the
contrary, a lot is going on behind the scenes, and we hope to be able to
offer many improvements to ZINC soon.

> 1) I tried looking for novobiocin, CAS 1476-53-5, NSC 2382, 
> in ZINC but did not find it. It is listed at the NCI/DTP 
> website.  How come?

We had fairly aggressive filtering critera that excluded numerous compounds
that probably should be in ZINC (and we intend to include in the next
update). Second, we didn't ever attempt to load every compound from every
supplier. In the future, we intend to load every "plausibly drug-like"
compound into ZINC from every supplier we work with. A third reason why
compounds didn't get loaded into ZINC was that our processing pipeline would
fail silently at various stages. We've completely re-engineered our molecule
processing pipeline. Our "dropout rate" in which compounds fail to reach the
end of the pipeline is a small fraction of what it was.

> 2) I still haven't been able to upload my own molecules (a 
> humble 20) into zinc to get them ready for docking. how are 
> the upgrades coming?

Would you write me (not the list please) with your molecules and I'll
investigate what's going on.

> I read sybyl is not recomended for small molecule charge 
> adding...should I use sybyl to get mol2 files with charges 
> from the PDB files I have? or is it worth the wait?

First you need to get mol2 from PDB. In general PDB to mol2 requires human
assistance to resolve problems. If you have only a few molecules, sybyl
should be fine. Just be prepared to spend a minute or two cleaning up the
atom types and bond orders. But you don't want charges from Sybyl, I think.
Use sybyl to clean up the mol2, but then use semiempirical or better (e.g.
am1-bcc) to get charges.
> 3) may I suggest that keeping subsets available for a whole 
> week is really a long time and that few days (3 or 4) should 
> be enough for anybody to download them? wouldn't you increase 
> the turnover time between sets by decreasing the time they 
> are stored? or even better, can you configure/develop the 
> software to make the computer smart enough to automatically 
> delete files as soon as they are downloaded?
> anyway, I hope things are working out over there.

The whole subset system needs re-working, which I will get to after the new
ZINC release.


> thanks
> jose
> _______________________________________________
> Zinc-fans mailing list
> Zinc-fans at docking.org
> http://blur.compbio.ucsf.edu/mailman/listinfo/zinc-fans

More information about the Zinc-fans mailing list