Chemical Drawing Tool Requirements

[mattodd]

Most people in the OSM consortium need, at some point, to draw molecules. Typical uses of the drawings are i) conversion to pictures for sharing (e.g. in Github posts, papers), ii) analysis of the structure in some way using other tools or iii) extraction of strings such as SMILES.

What do people typically use for this?

Folks at the University of Sydney have Chemdraw. Other unis may be the same. I think others use Chemsketch? What else?

Is there a need for another solution? Perhaps access to a solution for all contributors to OSM?

I ask in part because I've been contacted by ChemAxon to talk about our needs. @cdsouthan has mentioned these guys in the past. As it stands, on the demo page, one can play with MarvinJS and use it to save molecules as pictures like this one

or export the molecule in any number of ways (SMILES, .cdx, mol etc). People could continuously use the demo page to draw ad hoc molecules (!) but I'm assuming that acquiring access to the actual software might be more useful. What's the ideal outcome here? That we could as a group gain access to an online drawing tool (no requirement for software download) that allows drawing/saving/sharing of structures? Open source software would of course be good, but I see this (for tools) as less important than the use of something that is effective.

[cdsouthan]

Let me just start with an old chestnut comment about OS vs commercial (CS) for OSDD. As @mattodd says utility/robustness ranks highly if costs can be managed. Stating the obvious the I/Os for any CS can be set free as a bird and be totally open, but we just can't hack their code base. Equally obvious many OSM project collaborators come from Unis where ChemDraw, Microsoft, and SciFinder are already costed into departmental overheads and default/free at local source (but JFTR Edin does not have ChemDraw!). Equally obvious some lucky folk have licenses to expensive CS that could be high value to apply (and publish) within the project (e.g. Cresset). Note also we can very usefully use free stuff where never see the CS source code (e.g. Google Docs, DataWarrior, Chemicalize and even the back-end functionality inside PubChem and SureChEMBL).

As a direct comment on the question, the ubiquity of Marvin across public databases, including as default drawing input for ChEMBL, SureChEMBL, BindingDB, and (our own) Guide to Pharmacology, independently indicates it has something going for it. Its also well supported and I'm guessing Syd Uni-wide licensing of this and other ChemAxon stuff should not be too onerous. Note also you could get JChem for Excel and last but not least, even contemplate their chemical registration system....

Declaration of interest: some moons ago I was invited to a ChemAxon user meeting https://www.youtube.com/watch?v=OKLw9BaQzY0 (see comment at ~14 mins on CDX/Marvin) and I also wrote a paper with them https://www.ncbi.nlm.nih.gov/pubmed/23618056

[mattodd]

Thanks, @cdsouthan . Yes, there are two modalities here, if that's the right word. One is those of us contributing frequently and for whom software might be available through our institutions. The other is someone coming to the project for the first time from elsewhere, who would like to contribute but who may not have resources. e.g. a school student, a retired pharma scientist, anyone. If we're able it'd be nice to assemble a list of available resources (in a Tech Ops area) for such people to be able to contribute most easily - I think we were strongly behind DataWarrior in this way. If we could establish a way to use chemical drawing software like this, it'd help, I think.

One thing I’m particularly keen to seek is inter-operability with file formats. Thus it would help if, for chemical drawing, any tools we use can exchange files. Import, alter, export. I see this as more important than converging on a particular application. That Marvin can use .cdx and .mol is a big plus.

[drc007]

@mattodd @cdsouthan There are two distinct needs here, drawing structures for database searching and publication ready graphics. Unfortunately while some drawing packages excel in one or the other few (none?) really excel at both. ChemSpider uses Ketcher, Elemental and Accelrys Draw as the structure during options, PubChem have their own editor, ChEMBL use Marvin JS or JSME. Others use ChemDoodle web components. The javascript based tools on websites offer nice lightweight tools for database searching, but they don't really offer the functionality you might require for creating images for publications.

With regard to interoperability whilst many of the tools can open several file formats and tools like OpenBabel (http://openbabel.org/wiki/Main_Page) can interconvert 100's of different file formats you should be aware that once the structure has been pasted into a word processing package, presentation application or spreadsheet it is unlikely that you will be able to open it with a different drawing tools. Indeed you may find that different versions of the same drawing package can't open it further editing. So you need to ensure that you keep copies of any structures you have embedded in a publication.

You should also remember that the structure file formats are designed to contain chemical structural information only, so there is no guarantee that any annotations/arrows/headings/boxes/labels will be transferred to other file formats. File formats like .cdx are binary file formats so you can't even open it in a text editor as a last resort, .mol, .sdf even SMILES probably offer greater flexibility.

Sorry for the pessimistic tone, this is something I've struggled with for 30 years or more, I've spoken to all the software vendors and there is zero interest in interoperability.

[drc007]

I wrote a comparison of chemical drawing packages some time ago. http://www.macinchem.org/reviews/chem-draw-packages.php

A little out of date now but perhaps worth a read.

[drc007]

Also http://www.macinchem.org/reviews/chemdoodle7/chemdoodle7review.php now updated http://www.macinchem.org/blog/files/b76319d7900b8b674f8e85a396ac36cf-1951.php

http://www.macinchem.org/reviews/ChemAxon_review/marvin.php

http://www.macinchem.org/blog/files/3cbb99de8c0c345d3879a75bf46057cd-2051.php

[cdsouthan]

A slightly radical but obvious suggestion. By direct collaboration with an agile publisher for our forthcoming S4 paper (e.g. but not limited to F1000 research) we could contrive to juxtapose live PubChem links below each of the pictured final compounds (n.b. purified intermediates could have CIDs as well). At the risk of sounding heretical the necessity for high-res structure renderings is thereby diminished. NOBA the InChIs and SMILES from the paper and/or supp dat would render as paste-ins for any sketcher/search front-ends worth using. The less experienced won't need this for searching anyway since PubChem pre-computes 2D and 3D neighbours (see https://github.com/OpenSourceMalaria/OSM_To_Do_List/issues/528)

For the record, in a past publication (https://www.ncbi.nlm.nih.gov/pubmed/23506624) I managed to pull the PubChem .sdf files into Marvin, drop the image out and paste across to, you guessed it, MSWord.

Sure, its neither as crisp as a ChemDraw rendering nor as aesthetic as a ChemSpider one (if we were allowed colour). Notwithstanding would it not do?

[drc007]

For which journal is the publication intended, many journal have very strict requirements about the software that can be used.

[mattodd]

@drc007 I'll open a new Issue on the Series 4 paper ASAP - just finding the time to sort out a preliminary something related to file sharing. Then we can discuss target journal.

[mcoster]

Kudos to the ChemAxon team - they got in touch with me after the local rep talked to @mattodd (thanks Mat!) at RACI100. I mentioned to them that the aesthetics of MarvinSketch output is a stumbling block, and they pointed me to some features I wasn't aware of, including the ability to choose various journal styles: "File -> Document Style -> Settings -> Apply journal style preset". Below is an example using the "ACS" style:

There are still a few (mainly usability) quirks, but I'm impressed so far.

[cdsouthan]

Fine, AWAK there are big advantages to a good local rep, including "please fix this or that" or "c'mon that quoted price is outrageous-try again" without needing a TC with Hungary. NOBA I have mentioned to @mattodd before there must be other biological chemistry groups in Syd Uni with essentially identical informatics pain points who, if they all joined forces (a bit of cat herding?), could give collective leverage. I'm guessing Dotmatics as a possible alternative with a good reputation in this space, don't have a local presence?

[danormsby]

We (Dotmatics) have a guy in Melbourne. Elemental might suit your needs. Does export to MOL/RXN/SMILES and PNG images are easy to pull from the canvas object it sits in.

Cheers, Dan.

[cdsouthan]

Well - there you go @mattodd, a good chance of at least two free lunches then? :)

[madgpap]

Hey Dan, good to hear from you!

George.

Sent from my giPhone

On 8 Aug 2017, at 18:50, danormsby notifications@github.com wrote:

We (Dotmatics) have a guy in Melbourne. Elemental might suit your needs. Does export to MOL/RXN/SMILES and PNG images are easy to pull from the canvas object it sits in.

Cheers, Dan.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

[holeung]

I have been using ChemAxon software for several years. They provide free academic licenses. I highly recommend their software: easy to use, modern, and play nicely with other software.

[mattodd]

Thanks @holeung - am interested in whether OSM might be granted a shared licence that could cover occasional contributors. 5 concurrent users, for example.

[mattodd]

Had a call with ChemAxon today. They did a nice job of showing me demos of what OSM might be interested in.

The free demo version of Marvin JS I referred to above can be used, but close inspection of the license will, I’m told, reveal that we can’t use it in a repetitive, major way. i.e. if we intend to use it a lot, we would need to negotiate a license.

The way JS can be used is: the software is hosted on a server for us, and we obtain a static URL for Open Source Malaria. This could be used by anyone with the URL (I’m not sure if there would be a login requirement, but we could share that if so) but this web version of the software could be used by concurrent users to draw a molecule, get strings from it and make it into a picture. This seems like it’d be useful for ad hoc community users. Each user would have a blank canvas each time, which would be their own (i.e. not a shared canvas). Closing the browser window would typically erase the diagram, but there are ways to make that persist for each user, I’m told.

Seems useful as a go-to way for contributors to draw molecules. I’ve asked for a quote.

The fancier version is Marvin Live: also web-based, so also platform independent and needing no software download. Requires a login (can be a shared login), but one can make “rooms” where people can collaborate on diagrams. Snapshots can be taken at any time as a diagram progresses and those snapshots can be reverted to, or linked to in the future. i.e. a nice way to manage versioning for diagrams. The rooms cannot, however, be viewed without a login, which makes them of limited value to an open source project. As it stands one cannot search across all the molecules (e.g. by structure) that a given user has ever sketched, though this functionality could probably be developed.

[mcoster]

It's an interesting dilemma - an initiative like OSM doesn't seem to fit within the framework of most commercial licenses.

Not wanting to add to the confusion, but I haven't seen any mention of BKChem - free, open source chemical structure drawing software written in python. It hasn't seen much active development in recent years, I find the interface unintuitive and difficult and the project homepage has lots of dead links, but it is free, open-source and if you dive into it, you can do all the basics, plus output to high quality .svg files. I would never use it in preference to ChemDraw or even MarvinSketch, but my structural chemistry colleague, who is entirely Linux-based, uses it exclusively for his structure drawing needs.

Speaking of Open Source software, the above-mentioned colleague of mine has a handy tool for _c_ompound _Appraisal - [cApp_](http://www.structuralchemistry.org/pcsb/capp_home.php). I see that @drc007 even reviewed an earlier version in 2015!

[cdsouthan]

Have/will you kick the tyres on the new CDK @mcoster ?

[cdsouthan]

Oh wow, the real lowdown https://www.slideshare.net/NextMoveSoftware/rdkit-ugm-2016-higher-quality-chemical-depictions?next_slideshow=1

[drc007]

I might be worth be worth making a distinction between a chemoinformatics toolkit (that might be running on a server) and a desktop client. OpenBabel, RDKit and CDK are open-source toolkits that can be used to build applications. A couple of issues, long term support for an application built on one of these toolkits can be a problem if an update breaks something. OpenBabel and RDKit can be accessed using Python, CDK is java based, whilst Python is installed in all major desktop operating systems, Java is not installed and requires a separate install. Several java apps have now been rewritten in javascript and I suspect this may be the direction to follow for desktop/web apps in the future. I use all these toolkits regularly using scripts I've written, I'm not sure I'd use them for the basis of a chemical drawing package.

[kkruups]

Hi Folks,

I don't know specifics of drawing these molecules in the drawing programs mentioned above, but could you use Something like CanvasMol. : http://alteredqualia.com/canvasmol/#Aluminium_oxide

You'd can import an add or tab of file of molecule. Then manipulate the model and even expect to ping format image file. Something like Canvamol allows you select specific options to view, bonds, bonds and atoms, view atoms as element letter or colored sphere. I think the Code on Github and can be forked, so features can be changed or added. I also, think if you contact the owner he would be happy to help.

BR/Kk