3D optimizations using Rajarshi's optimizer now on ChemSpider.

13 views
Skip to first unread message

Tony at ChemSpider

unread,
Sep 24, 2007, 8:22:34 PM9/24/07
to UsefulChem
Thanks to Rajarshi for pointing us to smi23d we have now implemented
this code underneath JMol and when you try and display the structures
in Jmol now BEFORE they display they are optimized.

This is our beta implementation so we do not yet have a statement
saying "optimizing 3D structure" so for complex structures the JMol
screen stays blank while generating the 3D coordinates....so be it.

REad these blog postings from last night re. Diazonamide A...

http://www.chemspider.com/blog/?p=147
http://www.chemspider.com/blog/?p=153


Now, take a look at http://www.chemspider.com/Search.aspx?q=diazonamide+a
Look at the bottom 4 structures out of the 6 and if you click on Jmol
you will see the structures optimized. Click on the top 2 in Jmol and
the optimization fails. Rajarshi...we need your assistance please!

Rajarshi

unread,
Sep 24, 2007, 9:12:47 PM9/24/07
to UsefulChem

On Sep 24, 8:22 pm, Tony at ChemSpider <tony27...@gmail.com> wrote:
> Now, take a look athttp://www.chemspider.com/Search.aspx?q=diazonamide+a


> Look at the bottom 4 structures out of the 6 and if you click on Jmol
> you will see the structures optimized. Click on the top 2 in Jmol and
> the optimization fails. Rajarshi...we need your assistance please!

I just tried the first one locally and it seems to work fine. I took
the isomeric SMILES from the Pubchem CID 395475 (as linked to from the
Chemspider entry), ran it through smi23d (< 1 sec) and then mengine (<
10sec) and I was able to view the resultant SDF in Jmol locally.

I'm not sure what's going on - what SMILES are you providing to smi23d?

Val at ChemSpider

unread,
Sep 24, 2007, 9:19:29 PM9/24/07
to usefu...@googlegroups.com
Hi Rajarshi

We don't feed SMILES, we feed 2D SDF to mengine. This is the output I
get:

field : MMX
Atom Types: 169
Bonds: 580 Bond3: 0 Bond4: 0 Bond5: 0
Angle: 434 Angle3: 41 Angle4: 60 Angle5: 0
Torsion: 697 Torsion4: 58 Torsion5: 0
Vdw: 172 OOP: 91 Dipole: 474 Charge: 0 Improper: 0
STBN: 26 ANGANG: 0 STRTOR: 0 VDWPR: 4
field : MMFF94
Atom Types: 181
Bonds: 448 Bond3: 0 Bond4: 0 Bond5: 0
Angle: 1801 Angle3: 21 Angle4: 61 Angle5: 0
Torsion: 674 Torsion4: 38 Torsion5: 95
Vdw: 182 OOP: 112 Dipole: 0 Charge: 0 Improper: 0
STBN: 286 ANGANG: 0 STRTOR: 0 VDWPR: 0
Ngood: 0 Nbparam: 0 Nbatom: 0 Nbcontact: 1

Val at ChemSpider

unread,
Sep 24, 2007, 9:22:43 PM9/24/07
to usefu...@googlegroups.com
And here is SDF I feed, sorry...
350562.mol

Rajarshi

unread,
Sep 24, 2007, 9:20:28 PM9/24/07
to UsefulChem

On Sep 24, 9:19 pm, "Val at ChemSpider" <v...@chemspider.com> wrote:
> Hi Rajarshi
>
> We don't feed SMILES, we feed 2D SDF to mengine. This is the output I
> get:
>
> field : MMX
> Atom Types: 169
> Bonds: 580 Bond3: 0 Bond4: 0 Bond5: 0
> Angle: 434 Angle3: 41 Angle4: 60 Angle5: 0
> Torsion: 697 Torsion4: 58 Torsion5: 0
> Vdw: 172 OOP: 91 Dipole: 474 Charge: 0 Improper: 0
> STBN: 26 ANGANG: 0 STRTOR: 0 VDWPR: 4
> field : MMFF94
> Atom Types: 181
> Bonds: 448 Bond3: 0 Bond4: 0 Bond5: 0
> Angle: 1801 Angle3: 21 Angle4: 61 Angle5: 0
> Torsion: 674 Torsion4: 38 Torsion5: 95
> Vdw: 182 OOP: 112 Dipole: 0 Charge: 0 Improper: 0
> STBN: 286 ANGANG: 0 STRTOR: 0 VDWPR: 0
> Ngood: 0 Nbparam: 0 Nbatom: 0 Nbcontact: 1

^^^^
Aah, it looks like there is some van der Waal clash going on in the 2D
SDF.

In general, you should not provide a 2D SDF to mengine - you should
provide some form of 3D. Otherwise it's possible that the optimization
routine (BFGS IIRC) will barf on stuff such as everything being in a
plane and so on.

Is there a reason that you have to feed 2D SDF's? The role of smi23d
is to get a rough set of 3D very fast (< 1 sec for this case) and
mengine will work fine with that.

If nothing you could try converting your 2D SDF to a randomized 3D SDF
- but then you can't guarantee absence of van der Waals clashes.

I'd still suggest going from SMILES via smi23d

Rajarshi

unread,
Sep 24, 2007, 9:24:08 PM9/24/07
to UsefulChem

On Sep 24, 9:22 pm, "Val at ChemSpider" <v...@chemspider.com> wrote:
> And here is SDF I feed, sorry...

Aargh! My eyes! :)

No, this SDF has no chance of working with mengine - everything seems
to be overlapping.

If this is what you're doing for the other 4 structures, I can see why
they might work - I assume you're just transferring the 2D coordinates
to SD format. And the 2D depictions seem to indicate that they're
relatively well laid out - so no clashes etc. Which is why the lower 4
work.

Val at ChemSpider

unread,
Sep 24, 2007, 9:42:25 PM9/24/07
to usefu...@googlegroups.com
Ok. Will try this next!

> -----Original Message-----
> From: usefu...@googlegroups.com [mailto:usefu...@googlegroups.com]
On
> Behalf Of Rajarshi
> Sent: Monday, September 24, 2007 9:20 PM
> To: UsefulChem
> Subject: [UsefulChem] Re: 3D optimizations using Rajarshi's optimizer
now
> on ChemSpider.
>
>
>
>

Egon Willighagen

unread,
Sep 25, 2007, 1:01:56 AM9/25/07
to usefu...@googlegroups.com
Hi Val,

On 9/25/07, Val at ChemSpider <va...@chemspider.com> wrote:
> Ok. Will try this next!

Great to see this feature online. Please correct on the ChemSpider
website JMol into Jmol. Seems minor, but there is/has been another
(US-based) JMol project in addition to our favorit Jmol.org project.
Thanx!

Egon

--
----
http://chem-bla-ics.blogspot.com/

Val at ChemSpider

unread,
Sep 25, 2007, 1:18:41 AM9/25/07
to usefu...@googlegroups.com
Hi Egon

Done. Typo was in just one place..but the most prominent. :)

Re great features.. we want to use some CDK capabilities at ChemSpider.
Specifically to calculate some descriptors. Any recommendations what we
should start from? Also, ChemSpider is ASP.NET based... to facilitate
interfacing, do you have any webservice to get connected to CDK
capabilities?

Regards
Val

> -----Original Message-----
> From: usefu...@googlegroups.com [mailto:usefu...@googlegroups.com]
On
> Behalf Of Egon Willighagen
> Sent: Tuesday, September 25, 2007 1:02 AM
> To: usefu...@googlegroups.com
> Subject: [UsefulChem] Re: 3D optimizations using Rajarshi's optimizer
now
> on ChemSpider.
>
>

Egon Willighagen

unread,
Sep 25, 2007, 1:41:00 AM9/25/07
to usefu...@googlegroups.com
Hi Val,

On 9/25/07, Val at ChemSpider <va...@chemspider.com> wrote:
> Done. Typo was in just one place..but the most prominent. :)

Thanx!

> Re great features.. we want to use some CDK capabilities at ChemSpider.
> Specifically to calculate some descriptors. Any recommendations what we
> should start from? Also, ChemSpider is ASP.NET based... to facilitate
> interfacing, do you have any webservice to get connected to CDK
> capabilities?

Rajarshi's group has a number of webservices; I'm sure he'll respond.
Right now I am working on a couple of mass spectroscopy related,
CDK-based SOAP web services. Maybe even BioMoby, if I get that online.
Rajarshi, you are not using BioMoby, are you?

Rajarshi

unread,
Sep 25, 2007, 2:00:57 AM9/25/07
to UsefulChem
On 9/25/07, Val at ChemSpider <va...@chemspider.com> wrote:

> Re great features.. we want to use some CDK capabilities at ChemSpider.
> Specifically to calculate some descriptors.


Do you have specific descriptors in mind? We currently provide TPSA,
XlogP
and Surface Area as descriptors (though I should check that they
work). You
can get the WSDL at
http://rguha.ath.cx:8080/cdkws/services/Descriptors?wsdl and Javadocs
at
http://cheminfo.informatics.indiana.edu/~rguha/ws/cdkws/api/net/sf/ciccgrid/cdkws/Descriptors.html

You should ignore the getDescriptors method (whose goal is to evaluate
all
available descriptors for a molecule) as it hasn't been updated in a
while.

Any recommendations what we
> should start from? Also, ChemSpider is ASP.NET based... to facilitate
> interfacing, do you have any webservice to get connected to CDK
> capabilities?


We've put up a number of CDK functionality up as a webservice - some
of it
trivial, some of it useful (such as 2D structures - all our DB
interfaces
provide 2D depictions, without worrying about how I can interface CDK
with
PHP/Python etc, since we just call a WS with SMILES) See
http://cheminfo.informatics.indiana.edu/~rguha/code/java/cdkws/cdkws.html

I don't know ASP.NET, but I'd assume that you can connect via SOAP?
They're
all SOAP based web services

--
Rajarshi Guha

Rajarshi

unread,
Sep 25, 2007, 2:01:34 AM9/25/07
to UsefulChem
On 9/25/07, Egon Willighagen <egon.wil...@gmail.com> wrote:
>
>
> Rajarshi's group has a number of webservices;


We do indeed and we had the CDK descriptors as a web service, but it's
not
been updated for a long time. Clearly it's time to do so :)

Right now I am working on a couple of mass spectroscopy related,
> CDK-based SOAP web services. Maybe even BioMoby, if I get that online.
> Rajarshi, you are not using BioMoby, are you?


No


--
Rajarshi Guha

Jean-Claude Bradley

unread,
Sep 25, 2007, 6:25:46 AM9/25/07
to usefu...@googlegroups.com
Wow - getting a realistic conformation for any molecule from ChemSpider would be awesome.  It ties in nicely with the upgrade to Andrew Lang's molecule rezzer in Second Life.  Andy is using Rajarshi's web services directly from InChI (and soon SMILES) to create the 3D molecules in SL.
 
I looked at few simple structures on CS.  One seems to be good:
 
while another misses the sp3 hybridized carbon on the ring and draws it flat
--
Jean-Claude Bradley, Ph. D.
E-Learning Coordinator for the College of Arts and Sciences
Associate Professor of Chemistry
Drexel University

http://drexel-coas-elearning.blogspot.com
http://drexel-coas-talks-mp3-podcast.blogspot.com/
http://usefulchem.blogspot.com

Tony at ChemSpider

unread,
Sep 25, 2007, 7:42:39 AM9/25/07
to UsefulChem
Rajarshi,
Can you take a look at this "locally" on your system and comment on
whether this is an algorithm issue or a "starting point"
issue...thanks

On Sep 25, 6:25 am, "Jean-Claude Bradley"


<jeanclaude.brad...@gmail.com> wrote:
> Wow - getting a realistic conformation for any molecule from ChemSpider
> would be awesome. It ties in nicely with the upgrade to Andrew Lang's
> molecule rezzer in Second Life. Andy is using Rajarshi's web services
> directly from InChI (and soon SMILES) to create the 3D molecules in SL.
>
> I looked at few simple structures on CS. One seems to be good:http://www.chemspider.com/RecordView.aspx?id=27282
>
> while another misses the sp3 hybridized carbon on the ring and draws it flathttp://www.chemspider.com/RecordView.aspx?id=15948
>

> On 9/25/07, Rajarshi <rajarshi.g...@gmail.com> wrote:

Rajarshi

unread,
Sep 25, 2007, 8:26:23 AM9/25/07
to UsefulChem

On Sep 25, 7:42 am, Tony at ChemSpider <tony27...@gmail.com> wrote:
> Rajarshi,
> Can you take a look at this "locally" on your system and comment on
> whether this is an algorithm issue or a "starting point"
> issue...thanks

Looks like a starting point issue. I used smi23d to get rough 3D and
then minimized by mengine. The resultant SDF is at http://rguha.ath.cx/~rguha/nsc2071.sdf
which has the ring in the chair conformation

BTW, the latest code for mengine in trunk will, by default, not
evaluate the dipole moment and various vibrational data. You can use
command line params to get these values if desired

Rajarshi

unread,
Sep 25, 2007, 8:38:45 AM9/25/07
to UsefulChem

On Sep 25, 7:42 am, Tony at ChemSpider <tony27...@gmail.com> wrote:
> Rajarshi,
> Can you take a look at this "locally" on your system and comment on
> whether this is an algorithm issue or a "starting point"
> issue...thanks

BTW, given the recent threads on web services, you might be interested
in the 3D coordinate web services - send in a SMILES get back an SDF.

WSDL at http://rguha.ath.cx:8080/threed/services/Coord3D?wsdl and
Javadocs are at http://cheminfo.informatics.indiana.edu/~rguha/ws/threed/api/net/sf/ciccgrid/threed/Coord3D.html

Reply all
Reply to author
Forward
0 new messages