Send Rdkit-discuss mailing list submissions to
[email protected]
To subscribe or unsubscribe via the World Wide Web, visit
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
or, via email, send a message with subject or body 'help' to
[email protected]
You can reach the person managing the list at
[email protected]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Rdkit-discuss digest..."
Today's Topics:
1. Re: Swapping Bonds in 2D->3D conversion? (Marshall Levesque)
2. Re: Swapping Bonds in 2D->3D conversion? (Greg Landrum)
3. Re: Swapping Bonds in 2D->3D conversion? (Marshall Levesque)
4. Re: Compiling on Red Hat linux (George Oakman)
5. Re: Compiling on Red Hat linux (Greg Landrum)
6. Aromatic and conjugated bonds to singles and doubles
(George Oakman)
----------------------------------------------------------------------
Message: 1
Date: Mon, 30 Mar 2009 22:26:22 -0700
From: Marshall Levesque <[email protected]>
Subject: Re: [Rdkit-discuss] Swapping Bonds in 2D->3D conversion?
To: Greg Landrum <[email protected]>
Cc: [email protected]
Message-ID:
<[email protected]>
Content-Type: text/plain; charset=ISO-8859-1
Greg-
I gotta say that the "RDKit Tech Support Team" is phenomenal at
getting questions answered!
The previous spiderweb of a 3D structure made it not at all obvious to
me that a little more optimization would do the trick.
I will look into adding the additional UFFOptimizeMolecule calls.
Is there a rough estimate you have for the MAX number of attempts one
would make with a 'screening compound' sized molecule in Optimizing
via RDKit?
As always... Thanks a lot!
-Marshall
On Mon, Mar 30, 2009 at 10:13 PM, Greg Landrum <[email protected]> wrote:
Marshall,
On Tue, Mar 31, 2009 at 6:56 AM, Marshall Levesque
<[email protected]> wrote:
I found there are a few molecules that have undergone a crazy transformation
from 2D to 3D with RDKit in my batch conversion due to having a long carbon
chain.
Here is one example molecules on eMolecules.com:
http://www.emolecules.com/cgi-bin/more?vid=600149
I've attached the input (2D) and output (3D) SDFs.
Here are other examples that result in the same "messy" 3D structures:
http://www.emolecules.com/cgi-bin/more?vid=479405
http://www.emolecules.com/cgi-bin/more?vid=445711
Has anyone seen this type of problem before? ?Is there a way to fix the
"messy" conversion? ?Or a way to detect when this long carbon chain is
present and tell RDKit to not convert/output these molecules?
looks to me like the UFF optimization just didn't converge. Here's a
better version, generated as follows:
#---------------
[3]>>> m = Chem.MolFromMolFile('crazymol.2D.sdf')
[6]>>> mh = Chem.AddHs(m)
[7]>>> AllChem.EmbedMolecule(mh)
Out[7] 0
[8]>>> AllChem.UFFOptimizeMolecule(mh)
Out[8] 1
[9]>>> AllChem.UFFOptimizeMolecule(mh)
Out[9] 1
[10]>>> AllChem.UFFOptimizeMolecule(mh)
Out[10] 1
[11]>>> AllChem.UFFOptimizeMolecule(mh)
Out[11] 0
[12]>>> print >>file('crazymol.3D.gl.sdf','w+'),Chem.MolToMolBlock(mh)
#---------------
converting that series of calls to UFFOptimizeMolecule to something
general is left as an exercise to the reader. :-)
-greg
------------------------------
Message: 2
Date: Tue, 31 Mar 2009 07:43:07 +0200
From: Greg Landrum <[email protected]>
Subject: Re: [Rdkit-discuss] Swapping Bonds in 2D->3D conversion?
To: Marshall Levesque <[email protected]>
Cc: [email protected]
Message-ID:
<[email protected]>
Content-Type: text/plain; charset=ISO-8859-1
On Tue, Mar 31, 2009 at 7:26 AM, Marshall Levesque
<[email protected]> wrote:
Greg-
I gotta say that the "RDKit Tech Support Team" is phenomenal at
getting questions answered!
yeah, well, the theory is that after a while other people will be able
to answer questions as well. hint, hint. :-)
The previous spiderweb of a 3D structure made it not at all obvious to
me that a little more ?optimization would do the trick.
I will look into adding the additional UFFOptimizeMolecule calls.
Is there a rough estimate you have for the MAX number of attempts one
would make with a 'screening compound' sized molecule in Optimizing
via RDKit?
Not really. This isn't something I have spent much time investigating.
It's probably also true that the defaults are not particularly well
chosen. You may want to increase maxIters to 500 when calling
UFFOptimizeMolecule and only try once or twice. There is a slight
speed penalty in calling UFFOptimizeMolecule repeatedly instead of
doing it once with maxIters increased, but there may also be a win
because doing so resets the optimizer (which is neither clever nor
massively robust). I'm afraid you'll have to experiment.
-greg
------------------------------
Message: 3
Date: Mon, 30 Mar 2009 23:14:59 -0700
From: Marshall Levesque <[email protected]>
Subject: Re: [Rdkit-discuss] Swapping Bonds in 2D->3D conversion?
To: Greg Landrum <[email protected]>
Cc: [email protected]
Message-ID:
<[email protected]>
Content-Type: text/plain; charset=ISO-8859-1
Greg-
I spend too much time on the RDKit sourceforge and google code
pages... so I'm sure I will end up answering questions.
I put in the change to the batch process code... with a max of 1000
iterations, and 10 max attempts. Went through a couple thousand
molecules pretty quick. I think there were only a few molecules in
this test file that needed multiple attempts for optimization.
RDKit has been flying through this stuff so I don't think there will
be much of a slowdown.
Thanks!
-Marshall
On Mon, Mar 30, 2009 at 10:43 PM, Greg Landrum <[email protected]> wrote:
On Tue, Mar 31, 2009 at 7:26 AM, Marshall Levesque
<[email protected]> wrote:
Greg-
I gotta say that the "RDKit Tech Support Team" is phenomenal at
getting questions answered!
yeah, well, the theory is that after a while other people will be able
to answer questions as well. hint, hint. :-)
The previous spiderweb of a 3D structure made it not at all obvious to
me that a little more ?optimization would do the trick.
I will look into adding the additional UFFOptimizeMolecule calls.
Is there a rough estimate you have for the MAX number of attempts one
would make with a 'screening compound' sized molecule in Optimizing
via RDKit?
Not really. This isn't something I have spent much time investigating.
It's probably also true that the defaults are not particularly well
chosen. You may want to increase maxIters to 500 when calling
UFFOptimizeMolecule and only try once or twice. There is a slight
speed penalty in calling UFFOptimizeMolecule repeatedly instead of
doing it once with maxIters increased, but there may also be a win
because doing so resets the optimizer (which is neither clever nor
massively robust). I'm afraid you'll have to experiment.
-greg
------------------------------
Message: 4
Date: Wed, 1 Apr 2009 17:52:49 +0000
From: George Oakman <[email protected]>
Subject: Re: [Rdkit-discuss] Compiling on Red Hat linux
To: <[email protected]>
Message-ID: <[email protected]>
Content-Type: text/plain; charset="windows-1252"
Hi All,
Just to close this thread.
My initial problem was indeed related to the LAPACK and BLAS libraries being
compiled for 32-bit while I needed them to be for 64-bit so that I can use the
address-model=64 mode.
The 64-bit versions of the library were part of the RHEL4 distribution files so
simply installing the pre-complied libraries from the distribution solved the
problem. As simple as that!
I can now compile the C++ libraries on RHEL4 with no errors.
Thanks,
George.
From: [email protected]
To: [email protected]
Date: Fri, 27 Mar 2009 20:36:36 +0000
CC: [email protected]
Subject: Re: [Rdkit-discuss] Compiling on Red Hat linux
Thank you all for great comments, what I need to so is starting to be clearer.
I just can't find a X86_64 lapack RPM for RHEL4, but I'll keep looking. If one of you knows where I could find it that'd be great.
George.
From: [email protected]
To: [email protected]
Date: Fri, 27 Mar 2009 15:46:10 -0400
CC: [email protected]
Subject: Re: [Rdkit-discuss] Compiling on Red Hat linux
How about now?
http://code.google.com/p/rdkit/wiki/BuildingOnLinux
or
http://code.google.com/p/rdkit/wiki/NewLinuxBuild
Ah, this is much better! Would it be possible to add a bullet point with
options for python-less build?
even (?!) the default?
That's definitely possible, but I wonder how advisable it is.
What fraction of active linux boxes are running 64bit?
I switched all of the linux computers I control to 64 bit for a couple
of years now. Out of 60+ systems I think I have 32-bit installed on 2,
one being my old home PC...
There is simply no good reason to run 32-bit linux on 64-bit hardware.
All of the old 32-bit programs work fine for me, and there is a bonus
that I don't have the ancient 2/4 Gb filesize/RAM limits.
Igor
------------------------------------------------------------------------------
_______________________________________________
Rdkit-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Windows Live Messenger just got better. Find out more!
_________________________________________________________________
Beyond Hotmail ? see what else you can do with Windows Live.
http://clk.atdmt.com/UKM/go/134665375/direct/01/
-------------- next part --------------
An HTML attachment was scrubbed...
------------------------------
Message: 5
Date: Wed, 1 Apr 2009 20:04:26 +0200
From: Greg Landrum <[email protected]>
Subject: Re: [Rdkit-discuss] Compiling on Red Hat linux
To: George Oakman <[email protected]>
Cc: [email protected]
Message-ID:
<[email protected]>
Content-Type: text/plain; charset=ISO-8859-1
Dear George,
On Wed, Apr 1, 2009 at 7:52 PM, George Oakman <[email protected]> wrote:
Just to close this thread.
My initial problem was indeed?related to?the?LAPACK?and BLAS libraries being
compiled for 32-bit?while I needed them to be?for 64-bit so that I can use
the address-model=64 mode.
The 64-bit versions of the library were part of the?RHEL4 distribution
files?so simply installing the pre-complied libraries from the distribution
solved the?problem. As simple as that!
I can now compile the C++ libraries on RHEL4?with no errors.
Glad to hear that you got things working.
Good luck with the code.
-greg
------------------------------
Message: 6
Date: Wed, 1 Apr 2009 18:10:15 +0000
From: George Oakman <[email protected]>
Subject: [Rdkit-discuss] Aromatic and conjugated bonds to singles and
doubles
To: <[email protected]>
Message-ID: <[email protected]>
Content-Type: text/plain; charset="windows-1252"
Hi All,
I would like to explore the possiblilty of transforming aromatic bonds into
single and double bonds, for example:
c1ccccc1 into C1=CC=CC=C1 into
and similarly, transform conjugated bonds into single and doubles, for example:
N:C:O into NC=O
Is there a utility within the RDKit that would help do something like this?
Thanks,
George.
_________________________________________________________________
Share your photos with Windows Live Photos ? Free.
http://clk.atdmt.com/UKM/go/134665338/direct/01/
-------------- next part --------------
An HTML attachment was scrubbed...
------------------------------
------------------------------------------------------------------------------
------------------------------
_______________________________________________
Rdkit-discuss mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
End of Rdkit-discuss Digest, Vol 18, Issue 1
********************************************
