The Fall ACS meeting, held in Washington DC, is over and I can get back to breathing. The CINF program at the meeting was excellent (but I’m biased) with a variety of symposia. Notable amongst them was the Herman Skolnik symposium, honoring this years Herman Skolnik awardee, Yvonne Martin. Of all the talks, two that caught my eye. Jonathan Goodmans’ talk showed the first case of an InChI key collision, which appears to have been confirmed by ChemSpider and Symyx. The second talk was by Anthony Nicholls of OpenEye at the Herman Skolnik symposium. He discussed the issue of ‘hyperparametric modeling’, i.e., the fact that many types of models ranging from QSAR to molecular dynamics and ab initio methods are over paarmetrized. While this problem is very obvious in QSAR modeling (hence feature selection), the implicit parametrization in things like force fields, ab initio methods is not always obvious. He also discussed issues with model validation techniques such as cross-validation and y-scrambling and suggested techniques that might be better to assess a model. In essence he suggests that we will be more rigorous about model quality when we start assesing the “risk” of a bad model, rather than just how well the model fits the data (which can obviously be misleading). While these ideas are not brand new, the examples and the presentation were eye opening. The last talk in General Papers on Thursday was also quite interesting – B.-Y. Jin spoke about topological rules defining the structures of various nanotube and fullerene structures and how he had been constructing them as bead models. Examples of them can be found on his blog .
I was also involved in organizing a symposium on scripting and programming with COMP. We initially feared that the topic might be too geeky, but we were pleasntly surprised by the size of the audience. We had speakers talking about their favorite languages and how they do modeling and cheminformatics using them. I spoke about R and rcdk – which turned out to be so hot that we were interrupted by a fire alarm (three times!)
On the social events side of things, it was good to meet up with old friends and make new ones. CINF hosted a number of receptions, all excellent. A dinner for CINF functionaries was held at Zaytinya – a Mediterranean tapas restaurant. All I can say is: mindblowing! I had never realized that shrimp and dill could be such a killer combination. The Blue Obelisk dinner was held at PS7, serving eclectic American cuisine. The food was nice, though the service wasn’t too great.
But I have a few days to rest, after which I’ll get on to the San Francisco program (which I must say, is looking very cool)