User talk:Frodo: Difference between revisions

From DISI
Jump to navigation Jump to search
No edit summary
No edit summary
Line 1: Line 1:
This is the Frodo talk page. Notes to myself.
This is the Frodo talk page. Notes to myself.


== ZINC ==
* broken -  http://zinc12.docking.org/substance/3850657
* C04798508 and C13952160 are the same. many more.
* metal detection in dock blaster. correct treatment of hi library
* 17005395 has charge -1 and  6565555 has charge -1


== Blaster ==
From Oliv: Here are some comments on Dockblaster, which I received as feedback from non-experienced users. It seems that the results page is hard to interpret:
- it is not clear to most people what the 4 schemes (coarse/finer/polarized/Amber) mean and what they should use for docking
- the bug that rank and rmsds do not appear in most cases is confusing. Rank and rmsds should also be labeled.
- Instead of "Browse subset 12" one could write "Click here for a list of the best 100 molecules"
- The results (what people want!) are well hidden at the bottom of the page. Maybe one could give the people only the rec.crg file, eel1/mol2 files? At a more prominent place of the page?
SARAH: hi john- just noticed that indole in the l99a binders set looked funny, so i checked the mol2 files out and ref1 is incorrect; the double bond is in the wrong place. ref2 is the correct form of the molecule. the zinc id is C14516984.
I've had a few goes now with Dockblaster on a couple of targets but am having problems with some Calcium-dependant phospholipases in which the catalytic Ca++ is not being included in rec.pdb .  First up I used a couple of structures from our group (e.g. I kqu 1j1a,) but in both cases the calcium doesn't get included. I then tried to prep the protein myself and submit those to Dockblaster but the same problem recurred.  Thinking it might just be our structures just this morning I used some other workers PLA2 structure (1DB5 Dockblaster ID #38751) and it also doesn't include the Ca. By comparison a test run with thermolysin (5TLN Dockblaster run # #38752) included it's
Ryan writes:  I'm finding many groups that are +2 when Marvin is sure they are 0,
maybe sometimes +1. I'm including SMILES strings which hopefully are
helpful to see what I'm talking about. I think part of the problem, or
maybe the reason things are okay and I'm crazy, is that when you look
at the ZINC page
http://zinc12.docking.org/substance/07667606
The benzimidazole group is drawn with all single bonds on the 5-ring,
but when you copy/paste the smiles into Marvin or another program
there are some double bonds. Very strange (or I'm possibly just
confused by chemistry again)
O=C1N=[C]2=[C](C=CC=C2)=N1
Some example ZINC codes follow...
07667606
10201533
57765079
01805265
34781275
Another group that seems to have a similar issue:
CN=C1N=[C]2=[C](C=CC=C2)=N1
12323863
17304169
13688743
11635241
20744559
I have an entire docking run that seems to be full of these 2 if you
need more examples.
= blaster =
dhi, 3frj, /raid3/people/mysinger/pxc/dude/species/dhi1/P28845/blasti/3frj/dock
NAP cofactor decapitated
hivpr, 1upj, /raid3/people/mysinger/pxc/dude/species/hivpr/Q72874/blasti/1upj/dock
asymmetric only half the biological, two ASP are 2.5A apart and both protonated
mtor, /raid3/people/mysinger/pxc/dude/species/mtor/P42345/blasti/*/dock
All ligands failed to build, so 0/5 were build automatically
-- metacyc update
http://bioinformatics.ai.sri.com/ptools/flatfile-samples/compounds.dat
-- pubchem update (quarterly)
-- cdd update (quarterly)
https://s3.amazonaws.com/cdd-public/cdd-public-structures-20110219230015.sdf.gz
For future downloads, you can visit
https://s3.amazonaws.com/cdd-public/ and figure out the current
filename from the XML that is displayed.


[[Category:Info]]
[[Category:Info]]

Revision as of 17:41, 18 March 2014

This is the Frodo talk page. Notes to myself.

ZINC

  • broken - http://zinc12.docking.org/substance/3850657
  • C04798508 and C13952160 are the same. many more.
  • metal detection in dock blaster. correct treatment of hi library
  • 17005395 has charge -1 and 6565555 has charge -1

Blaster

From Oliv: Here are some comments on Dockblaster, which I received as feedback from non-experienced users. It seems that the results page is hard to interpret:

- it is not clear to most people what the 4 schemes (coarse/finer/polarized/Amber) mean and what they should use for docking
- the bug that rank and rmsds do not appear in most cases is confusing. Rank and rmsds should also be labeled.
- Instead of "Browse subset 12" one could write "Click here for a list of the best 100 molecules"
- The results (what people want!) are well hidden at the bottom of the page. Maybe one could give the people only the rec.crg file, eel1/mol2 files? At a more prominent place of the page?


SARAH: hi john- just noticed that indole in the l99a binders set looked funny, so i checked the mol2 files out and ref1 is incorrect; the double bond is in the wrong place. ref2 is the correct form of the molecule. the zinc id is C14516984.

I've had a few goes now with Dockblaster on a couple of targets but am having problems with some Calcium-dependant phospholipases in which the catalytic Ca++ is not being included in rec.pdb . First up I used a couple of structures from our group (e.g. I kqu 1j1a,) but in both cases the calcium doesn't get included. I then tried to prep the protein myself and submit those to Dockblaster but the same problem recurred. Thinking it might just be our structures just this morning I used some other workers PLA2 structure (1DB5 Dockblaster ID #38751) and it also doesn't include the Ca. By comparison a test run with thermolysin (5TLN Dockblaster run # #38752) included it's

Ryan writes: I'm finding many groups that are +2 when Marvin is sure they are 0, maybe sometimes +1. I'm including SMILES strings which hopefully are helpful to see what I'm talking about. I think part of the problem, or maybe the reason things are okay and I'm crazy, is that when you look at the ZINC page

http://zinc12.docking.org/substance/07667606

The benzimidazole group is drawn with all single bonds on the 5-ring, but when you copy/paste the smiles into Marvin or another program there are some double bonds. Very strange (or I'm possibly just confused by chemistry again)

O=C1N=[C]2=[C](C=CC=C2)=N1

Some example ZINC codes follow...

07667606 10201533 57765079 01805265 34781275


Another group that seems to have a similar issue:

CN=C1N=[C]2=[C](C=CC=C2)=N1

12323863 17304169 13688743 11635241 20744559

I have an entire docking run that seems to be full of these 2 if you need more examples.

blaster

dhi, 3frj, /raid3/people/mysinger/pxc/dude/species/dhi1/P28845/blasti/3frj/dock NAP cofactor decapitated

hivpr, 1upj, /raid3/people/mysinger/pxc/dude/species/hivpr/Q72874/blasti/1upj/dock asymmetric only half the biological, two ASP are 2.5A apart and both protonated

mtor, /raid3/people/mysinger/pxc/dude/species/mtor/P42345/blasti/*/dock All ligands failed to build, so 0/5 were build automatically


-- metacyc update http://bioinformatics.ai.sri.com/ptools/flatfile-samples/compounds.dat


-- pubchem update (quarterly)


-- cdd update (quarterly)

https://s3.amazonaws.com/cdd-public/cdd-public-structures-20110219230015.sdf.gz

For future downloads, you can visit https://s3.amazonaws.com/cdd-public/ and figure out the current filename from the XML that is displayed.