[Dock-fans] Further analysis of the error messages when docking ZINC-derived library

Silberstein, Michael mikesilb at bnl.gov
Wed Jul 11 07:16:59 PDT 2007

Hi again,
This post is highly related to the last post that I made "Error messages when docking specific molecules withinZINC-derived library" and has more detail as to my current difficulties.
In the last post, I mentioned that many compounds in my ZINC database-derived library had errors of the form 
WARNING read_mol2: Error reading ATOM record for ZINC00002585 in ./xyz.mol2.
WARNING read_mol2: Error reading ATOM record for ZINC00002588 in ./xyz.mol2.
WARNING read_mol2: Error reading ATOM record for ZINC00002819 in ./xyz.mol2.
WARNING read_mol2: Improper BOND origin/target for ZINC00003421 in ./xyz.mol2.
and so on...
In this post, I will describe some highly odd results pertaining to one particular molecule that has been giving me the above error message.
For one of the compounds in my mini-library, ZINC00014272, I received the above error message, and started further investigating why this particular molecule was not working.
This molecule apparently has three separate .mol2 files.  (By the way, what dictates whether a given molecule might have multiple .mol2, SDF, SMILES (etc.) file formats??  Various protonation states?  Different conformers?  Or simply, different programs that generate various .mol2 outputs?  Why would ZINC00014272 contain three sub-entries?)
Using these three different .mol2 files, I noticed that in my mini-library docking output (using DOCK 4.0.1), only the 2nd of the three molecules provided docking output (i.e. molecules #1 and #3 provided the above error message).
I was curious as to why only #2 worked while #1 and #3 failed, so I started docking each of them individually and afterwards, in various combinations.  I received highly odd results.
The following represents the results that I got.  Maybe one of you can make some sense of it:
1. Prior 3000 member library docking - #2 docked followed by an error message (presumably from molecule #3)
2. Docked only #1 - Docking failed and program crashed  (mutilple computer address symbols showed up in the DOCK results.  If you want further clarification as to this 'crashed output', I can show it to you in a subsequent post.)
3. Docked only #2 - Docking failed and program crashed. (I found this to be surprising since in my larger library, #2 was the only one that docked normally.)
4. Docked only #3 - Docking failed and program crashed.
5. Docked #1 followed by #2 - #2 docked fine, although #1 had no error message whatsoever.  There were no error messages at all in the output.
6. Docked #1 followed by #3 - Docking failed and program crashed
7. Docked #2 followed by #3 - Docking failed and program crashed (Again, I found this quite surprising.)  One error message was found in output.  (My guess is that this is from #3)
8. Docked #2 followed by #1 - Docking failed and program crashed
9. Docked #1 followed by #2 followed by #3 - Exact same results as seen when docking the 3000-membered library.  Only #2 docked properly, while #3 (presumably) gave the error message. 
10. ZINC00012908 followed by #1 followed by #2 followed by #3 followed by ZINC00015788 - Same results as seen in scenario 9 (and scenario 1).  However surprisingly, the two flanking molecules ZINC00012908 and ZINC00015788 had no noticeable output whatsoever, and was clearly not docked at all.  It was as if these two molecules did not exist in the input .mol2 file at all.
So these are my results, and I find myself incredibly confused as to why only #2 worked, why it had to work only following #1, and why the additional (flanking) compounds were not even found within the DOCK output and were certainly not docked.
Again, this analysis was all performed using DOCK 4.0.1 .  Perhaps switching to a newer version of DOCK might help too.  However, I have already analyzed much of the results (the molecules that DID dock properly), and would rather focus on what I have, rather than to install the newer DOCK version.  I would again appreciate any comments, focusing on the DOCK 4 output, rather than simply shifting immediately to DOCK 5 and/or 6.  
If anyone can provide any insight as to these highly peculiar results, I would absolutely appreciate it.  Thank you very much in advance.
Mike Silberstein 

More information about the Dock-fans mailing list