Protein-Ligand docking talk

Here are the slides for the talk I gave last Friday on Protein-Ligand Docking at the EBI Small-Molecule Bioactivites course.

For three or so slides I drew heavily from Leach and Gillet's "Introduction to Cheminformatics", which I recommended to the course participants.

To follow the slides, I prepared a self-guided tutorial that used the GOLD docking software to dock ligands taken from Kinase SARfari to a kinase crystal structure taken from the PDB. If this is of interest, let me know.

EBI Small-molecule Bioactivities Course

I'm just back from an EBI workshop on Small Molecule Bioactivities where I was invited to talk on protein-ligand docking. The workshop was mainly focused on introducing the participants to the databases being maintained and developed by the ChEMBL team at the EBI. This team is lead by John Overington who blogs over at the ChEMBL-og.

The ChEMBL databases (ChEMBLdb and Kinase SARfari) are well worth checking out, and they have a very active team working on additional databases right now including one on SAR data for GPCRs, and another with information on clinical candidates. Apparently, the databases are designed to be easy to integrate with existing in-house proprietary data and this has already been done at the Institute for Cancer Research. At some point, they will be releasing a public interface to some of this data called CanSAR.

One interesting feature of this meeting was the number of systems biology people present, rather than hard-core cheminformaticians. These were PhD students and postdocs trying to integrate information from many different sources including small-molecule information.

I thought it was a great meeting, and I will be keeping an eye out for similar workshops in future. Word on the street is that Rajarshi will be over in mid-May to present a workshop on Cheminformatics in R so I might be back here again soon. That will cover RCDK and RPubChem among other topics (see links in an earlier blog post). I'll add a link here when registration opens.

Invitation to apply for a PhD in Cheminformatics

(Please forward to anyone who might be interested)

If you are interested in a PhD developing open source tools for cheminformatics, I would be interested to hear from you. The Irish Research Council has just opened this year's call for PhD applications for PhDs starting next September. This competitive funding is open to EU students only and is offered on the basis of the student's merit.

The student will be based in the School of Pharmacy, University College Cork. The exact project will depend on the background and interest of the student but will involve the development of software or algorithms in the area of cheminformatics. I have a track record of involvement in international collaborations on open source software such as OpenBabel, as well as developing novel algorithms in cheminformatics, and scoring functions for protein-ligand docking.

If interested, please send your CV along with a cover letter addressed to Dr. Noel O'Boyle at baoilleach@gmail.com. The cover letter should briefly describe why you are interested in applying for a PhD with me. Note that the IRCSET deadline is 10th February so applicants should get in touch as soon as possible.

Links: [My staff web page] [My research] [Noel O'Blog]

University College Cork: University College Cork, founded in 1845, is one of Ireland's oldest institutions of higher education. Its first professor of mathematics was George Boole who developed the logic that underlies modern computing. The university has 18,860 full-time students: almost 12,000 undergraduate students and some 3,200 students at postgraduate level.

Cork: Cork is situated in the south-west of Ireland, and is the second city of the Republic of Ireland. It was recently named by Lonely Planet as one of the top ten cities to visit in 2010. It is situated in County Cork, an area of outstanding natural beauty.

Chemical Identifier Resolver + TwirlyMol = Easily add molecules to a webpage II

Noticed anything different about the TwirlyMols over at Carbon-based Curiosities, Chemical Quantum Images, or Node in the Noosphere? (Here's a hint - perhaps they look more awesome?)

Yes, indeed, TwirlyMol now uses proper spherical shading for the atoms. In addition, if you are using any of the 5 or so browsers that are not Internet Explorer (shame on you, IE), you will see that distant atoms are now wreathed in a fog (really it's a representation of the uniform electron gas). Through the magic of the internets, any TwirlyMol using the Chemical Identifier Resolver will suddenly look a whole lot better.

We've also addressed some minor bugs: shadows in wrong place, loading issues with multiple molecules, slow loading, issues with Opera.

Markus has also been doing some work on caching. To take advantage of this new work (and shave several seconds off load times), use the following new HTML:

<div id="DIVNAME" height="200" width="200"></div>

<script src="http://cactus.nci.nih.gov/chemical/structure/
             CHEMICAL_IDENTIFIER/twirl_cached/DIVNAME"
        type="text/javascript"></script>

Here's an example of it in action, where the CHEMICAL_IDENTIFIER is restasis and the DIVNAME is foggymol:

Cheminformatics Tutorial using Python and Silverlight II

Part I

I've updated the Interactive Cheminformatics Tutorial by adding sections on descriptors and molecular fingerprints, and removing the original Python tutorial (following Michael Foord's suggestion). In summary, the tutorial could now form the basis of a course on cheminformatics.

Currently the tutorial focuses on the possibilities of the Cinfony API but a possible improvement would be to include more didactic material, for example describing SMILES strings, their syntax and so forth.

One of the nice things about the Chemical Identifier Resolver is that it allows you to reference molecules by name which really brings a bit more life to the examples. Using this I've tried to think up interesting examples (e.g. "What's the Tanimoto similarity of aspirin to Dr. Scholl's Wart Remover?"). Additional suggestions are very welcome.

Chemical Identifier Resolver + TwirlyMol = Easily add molecules to a webpage

Markus Sitzmann of the NCI/CADD team has been busy. He has combined the Chemical Identifier Resolver with TwirlyMol to enable you to convert any chemical identifier to a 3D model that can you interact with in your webpage. I'm very excited about this as I think that people will find this very useful.

Just put this in your webpage or blog post (note however that Blogger preview does not show the Twirlymol):

<div id="DIVNAME" height="200" width="200"></div>

<script src="http://cactus.nci.nih.gov/chemical/structure/
             CHEMICAL_IDENTIFIER/twirl?div_id=DIVNAME"
        type="text/javascript"></script>

Replace DIVNAME with a unique name, and replace CHEMICAL_IDENTIFIER with any of the chemical identifiers accepted by the Chemical Structure Resolver; for example, a common name for a chemical, an InChI, or a SMILES string. More details over at the /Chemical/Structure blog. For now, let's just see it in action.

Replacing DIVNAME with 'buckyball' and CHEMICAL_IDENTIFIER with 'buckminsterfullerene' gives the following (go on, give it a twirl! - right mouse button to zoom in):

That was too easy - let's take one of Henzy Rzepa's crazy Mobius aromatic molecules. Steven Bachrach has written a review of some and very thoughtfully has included the InChIs. Replacing DIVNAME with 'crazymolecule' and CHEMICAL_IDENTIFIER with 'InChI=1/C14H14/c1-2-6-10-14-12-8-4-3-7-11-13(14)9-5-1/h1-14H/b2-1-,4-3-,9-5-,10-6-,11-7-,12-8-/t13-,14+' we have:

3D Nanoputians anyone? Here's the SMILES for a NanoKid: "c1(C2OCCO2)c(C#CC(C)(C)C)cc(C#Cc2cc(C#CCCC)cc(C#CCCC)c2)c(C#CC(C)(C)C)c1". Happy twirling.

My beaker overfloweth - New chemistry Q&A sites

Stackoverflow is one of the best Question & Answer websites for computer programming. It uses a carefully designed social model to build a community where people compete to give the best answer to questions in order to be rewarded with a better response to their own questions.

Recently, the people behind Stackoverflow have opened up the software to allow people to set up their own websites...but just for a beta period (money will then be required). Several chemistry 'stackoverflows' have already been set up. Here are a few I've heard about:

BlueObelisk: Questions about cheminformatics and computational chemistry leaning towards the open source or open data side of things. Update (07/10/2010): This website has moved to Shapado.
Chempedia Lab: Questions about experimental chemistry.
Chemistry: General chemistry (?)

These sites are all new so you won't find many questions there already. But give them a go. Go there and ask a question or two (even if you already know the answer), answer a question or two, and check back in a day to see what happens. You can log in with your Gmail address (among others) but do note that questions are not anonymous.

Such websites require a community. Some will gain such a community and flourish, others won't and will fail. In the meanwhile, go get some answers.

Image credit: Question Everything (Nullius in verba) Take nobody's word for it by Duncan Hull (CC BY 2.0)