After trying this in MarvinSpace, it appears that MarvinSketch is the way to do this. Here is what I have done so far:
(1) Read in each 2D molecule into MarvinSpace, save it as 3D Mol. Do this for all six molecules.
(2) Read in each 3D molecule into MarvinSketch, and paste them all into one window.
(3) Align the molecules from the MarvinSketch window. If I do the default (flexible fit), it takes a long time--it might not even finish.
The problem with this is that it takes a long time, is restricted to a maximum of six molecules, plus if I want to add another molecule I would have to start over.
What I would like to do:
(1) Align two 3D molecules, selecting Molecule 1 which I assume means that this is fixed. This is vary fast.
(2) Save each--here is where I have trouble: there are two versions of each molecule, one CPK colored and one uniquely colored. Which should I pick?
(3) Bring in the original Molecule 1 and a third molecule and Align, again having Molecule 1 selected. Save Molecule 3.
The problem is that with this operation, Molecule 3 is not aligned on Molecules 1 and 2.
Is there a way around all of this? My molecules are not very flexible--the low energy conformer is fine for what I am doing.
You might like to consider doing this using KNIME when there are a number of third party tools around which can either fit pairwise or fit onto a pharmacophore etc.
Thanks for the suggestion, indeed, KNIME is a useful tool and most ChemAxon features are available in KNIME.
However, Yvonne asked not only about the pair-wise alignment but also about multi-molecule alignment which is not available in KNIME.
I admit that the alignment of several molecules in one go considering full conformational freedom of the structures takes long time. Is this limitation that made you consider to do it in a pair-wise manner? I mean, is your ultimate goal the multi-molecule alignment and the pair-wise approach is a compromise? Or the result that you can obtain from this latter approach would be as good and useful as with the simultaneous alignment?
We will build you an alignment tool that eliminates the difficulties you experienced when using MSketch. This will be made available in our sandbox, so you can experiment with it online without the need to download and install anything. Based on your feedback we can make quick modifications until the alignment tool meets your requirements.
The new alignment algorithm that I mentioned in Boston will be released in the the next major release 5.4 soon. The web-based tool that I outlined above will apply this newer method which operates faster than the one currently available in MSketch.
Would this web-based alignment tool be convenient for you? When its mature enough then we can integrate it in other tools, including MSketch, MSpace as well as workflow tools like KNIME.
The KNIME nodes needed relate to 3D conformer generation and 3D searching plus some means of generating/entering a fitting hypothesi.s Knowing Yvonne she will read between the lines and deduce that the KNIME extensions my company has developed have such nodes but I'd be surprised if those provided by main stream modelling companies did not also provide similar capabilities. In my experience the best approach is to find a common pharmacophore which tends to work well with diverse molecules but I'm well aware that traditional 2D thinking would focus instead on functional groups eg fitting based on maximum common substructure.
The great thing about KNIME is the concept of mix and match nodes from different vendors. This also releases each vendor from duplicating functionality that they know others can do better!
Indeed, thanks Keith for the useful comments.
The reason that I decided to try pair-wise is that I am working on a problem where I suspect that the main
issue is shape fit. In answer to your question, I think pair-wise would be most useful. I would need this for Step 2 below.
As to using the sandbox, I need to be sure that the site is secure--some of the molecules are confidential. I am nervous about sending them over the internet.
I have the following information and thoughts:
(1) I have an extensive SAR in a series of conformationally very
restricted molecules. Many substituents kill affinity and only one particular
substituent is required. I would like to build a “map” of this dataset to
highlight “forbidden” and “allowed “ regions.
(2) I also have a list of other active molecules, usually each has at least
one closely related inactive. Not all of these were tested in an identical
assay—for example, some are from patents. My thought was to superimpose each of
these on the best of Set 1. I suspect that not all of these will superimpose
identically—that there is no universal 3D shape pharmacophore. Rather, each
active occupies some part of the required space. Obviously, I would use this
information to suggest where to place substituents on which of these active
molecules to increase affinity.
all of this to work, I need the reference potent ligand to always remain at the
same coordinates and any molecules aligned with it to move to that coordinate
space. I suspect that this is a simple change to the code—is life ever simple?
Thank you Yvonne for the detailed explanation, it helped us a lot.
Based on these requirements we build a custom application for you to experiment with. It will not do anything else than superimpose a set of molecules, and it will enable to keep one reference structure unchanged (and in the exact initial position).
Is the order of pair-wise alignments significant? Or the set of structures can be superimposed on the reference structure in arbitrary order?
Regarding the security issues: this is manageable using Java WebStart technology. With this you access a website and the program is downloaded and installed on your machine in an automated way. In this process it is only this dedicated application that is downloaded and used, you existing Marvin and JChem installations are not affected. The alignment application runs on your machine, your structures or any other information does not leave you computer.
This technology also enables us to update the application any time, and whenever you try to use it always checks if there is a newer version available in the sandbox. If not, then the system proceeds with your recent version, otherwise it download and installs the newer one before launching the application locally.
Is this approach suitable for you?
The trial application sound like just what I was hoping for. Thanks so much for the offer. The security issues are resolved.
At the moment I don't think that the order of the superposition would be critical if the reference molecule is kept in its input conformation. Depending on the option chosen, the other molecules might adapt to that set of coordinates. In other cases I might just want to superimpose all (could be as many as 50) of the molecules over some core using just the Marvin minimum energy conformation.
I look forward to trying this out.
Adrian is progressing really well with the first quick and dirty prototype, it should be available soon. However, there might be some issues with Java Webstart on Mac OSX, and we'd like to make sure that you do not run into such problems when you first try the application.
I have a vague memory that you use a Mac, is that correct? In this case, can you tell us the version number? (Click the Apple icon in the top left corner, and choose the 'About this Mac' item from the pull down list.)
Thanks for the information,
I am using Mac OS X 10.5.8.
If it were necessary I could try this on my aging laptop. It would be less convenient, but I could do so.
Finally, the multi-molecule alignment demo application is available.
It took us longer time than usual to build it, as I mentioned previously, there were some Mac related issues. These have not yet been fully resolved, we had to make a workaround. This affects the visualisation part only, I believe this will not strongly influence the usability of the small application.
To start the program please click on this link: http://discoverygroup.chemaxon.com/
This will open the Discovery Tools Sandbox in a new Window. The third item in the list on the right hand side is the Alignment demo application. Click on this link to open another window with two variants of the same application, please choose the second one on your Mac. It starts the download and launches the application straight after the download has been completed.
The multi-molecule alignment application takes a rather simplistic approach in its first version, yet I hope it will make the functionality readily available for you. It is a raw implementation so please do not expect user friendly GUI with fancy features.
The aim is to let you an other interested users try the approach and thus facilitate a discussion between you (as well as other scientists) and us, developers.
Your feedback is very valuable for us: we are keen to learn your thoughts and we aim to develop you the most suitable application. We will try to modify/improve the alignment algorithm and implement new features in the program considering your requests and suggestions. The sandbox will enable us to update the program as often as needed, independently from the official ChemAxon releases.
Finally, please accept again my apologies for not being able to make this application for you sooner.
I will download the application in a moment and try it out today.
I see that my previous post didn't make it to the forum. I was using my iPad (), which didn't let me get to the message section. I will post a second comment to the forum. Although it would be handy for me when I travel, I am not sure that compatibility with and iPad is top priority.
As far as I can tell, because it isn't documented, one must read in all molecules at once, apparently in an sdf file. This is quite awkward because I don't know of an easy way to do this. The way that I tried to do so was to make an Instantjchem database by reading in my .mol files one at a time (losing the names) and saving them as an sdf. I don't know how to save only some of the files in InstantJchem. I could probably also use Marvin to convert generate the SMILES of these molecules and save that and the name in an Excel file to then cut-and-paste from that into a file to use as input.
I don't understand what the option "Try conformation" means. Is that how many conformations of the reference will be used?
(1) Allow the input of molecules in more than one step.
(2) All the user to signify certain molecules to be ignored.
(3) have an option to view the structures input as 2D even if 3D structures were read in or generated.
Beyond all this, I need a license file and hence I can't go further. It stops after it starts the alignment.
the iPad issue was forwarded to the web developper team I am sure they will find a fix soon.
Thank you for Your valuable feedback on the AlignmentGui. The following fixes and solutions are planned:
- Allow reading more sdf files in the molecule table and also allow removing some molecule rows permanently. There also will be a way to optionally save the edited list of input molecules in a new sdf file.
- One will be able to ignore/include selected molecules in the actual alignment run.
- In each row a 2D and a 3D molecule image will appear.
- Sales team will send you the missing licenses.
- "I don't understand what the option "Try conformation" means. Is that how many conformations of the reference will be used?"
Yes, exactly. The alignment algorithm uses 3-4 initial conformations to the flexible overlay. These conformations are generated on the fly without the need of any further things to do. Usage of more conformations results in nicer alignments with longer running times. (eg.some flexible molecules containing more rotatable bonds, like some thrombin inhibitors, may require 7 conformations but generally 3-4 is enough). A clarifying tooltip will also appear in the gui.
If you agree these solutions I will implement them in the next two/three days and let you know when it is ready to try.
Thank you again for reporting these issues.
the user interface update of the alignment is ready :
http://discoverygroup.chemaxon.com/ please follow the "Alignment GUI early preview" link.
An online video to demonstrate the operation of this alignment GUI will be posted here soon.
Feel free to try it and any further comments feature requests and bug reports are welcome.
The installation went fine and I was able to add and delete molecules from the list. However, when I tried to do the alignment a license file was required.
I apologize for being slow to test this--will watch for updates and get on it sooner.
I would like to be able to have a name with every molecule used. My sdf files come in with a name. However, I have a .mol file for which the only name is the file name.
I suggest that if there is no name in the file that the file name be the default name in the program display.
All names should be editable.
As far as I could understand you have some trouble around the alignment license. Is that right? Please help me to see this clear because there were several licenses circulating around in this topic. I've sent you a license on the 18th of November including the alignment too.
Do you have troubles with this one?
the licensing issue will be solved by the sales department. They will check the license file, which has been sent to you, and they will provide assistance for installing this file.
The original idea was if a molecule has a name field that will be shown. If there was no name found the IUPAC name is generated on the fly and that will be visible. If I understand correctly you suggest that this name field should be the filename instead of the IUPAC name like: myMolecules.mol > myMolecules#1, myMolecules#2 ....
Has the IUPAC name any extra benefit for you in this context?
Making name field editable is a good idea. I will implement it it soon.
Thank you for your valuable remarks and your time.
you have written:
"-will watch for updates and get on it sooner."
is that possible that you are not getting automatic notificaions to your email address when this post changes?
I am late in responding, I'm sorry.
As to the license issue, as I recall one was sent to me. However, I didn't know what to do with it. I couldn't find an older version of such a file on my computer, so didn't do anything further. Please send it to me again and tell me where I should put the file. I assume with the ChemAxon files, but where. If not, should it be somewhere else? Unfortunately, I am not at my Mac now so can't give more information.
As to names, I didn't see the IUPAC name on apomorphine. I would assume that most modellers would name their files with something to remind them what they contain, so giving the filename as an option for a molecule name would be worth doing. However, having it or the IUPAC name editable is the best solution, because sometime the filename contains information that would no longer be true once an alignment was performed.
I think that I am getting the notices of the forum, thanks.
I've just sent you an email with the current, working license file attached to that. Please use that one.
There's no special requirement where to place the license file itself. However, you have to install that! We have a guide on our site about installing ChemAxon licenses: http://www.chemaxon.com/marvin/help/licensedoc/installToDesktop.html#gui
I truly hope that this will solve your issues,
Have a nice day,
I am totally frustrated.
I ran the license manager and got the response that a lot of licenses were updated including one named "Alignment".
However, the alignment application still does not run. It is looking for "3D Alignment"--could that be the problem?
Instant JChem works OK, so some of the licenses are working.
sorry for Your inconvenience. I could reproduce this issue with your license but not with mine.
We are handling this issue high priority and I hope we will solve it soon.
the license issue is temporary solved though You may obtain a newer license file in the future.
The molecule fields are now editable after double click and may have the following default values:
- filename #molname : if read from file and no molecule name field was found
- molecule field name : if read from file and molecule name field was found
- IUPAC name : if user had drown a new molecule here
Any feedback is welcome.