On Wed, Mar 18, 2009 at 3:39 AM, Andrew Dalke <[email protected]> wrote: > On Mar 12, 2009, at 5:49 PM, Greg Landrum wrote: >> The thing I don't know how to do, and that I suspect would be >> considerably >> more effort, but that would also be very cool, would be to add >> chemistry functionality to Excel (like what chemfinder for Excel >> does). > > It's possible. I did a prototype like that for a client, but it > didn't go anywhere. In my case it was the ability to have a column of > SMILES, write the headers of the descriptors you wanted, and the add- > in would go to a XML-RPC server to fill in the missing values. The > code used Twisted and wxPython to do the networking and GUI > components in Excel. > > The key resource for how I did it is the SpamBayes add-in for > Outlook, at > > http://spambayes.sourceforge.net/ > > This is an Outlook extension, but the API for how to do add-ins > (extensions) to Outlook was similar to that for Excel. > > I would be interested in working on it further except 1) I wasn't > convinced that it would be profitable for me and 2) I don't have the > right background to know what's useful, so I would need to > collaborate with an actual end-user so the result is actually usable > and enjoyable.
Years ago (the files say 2001) I did a prototype of some COM pieces for Excel from Python (the code is actually in the RDKit: $RDBASE/rdkit/ML/Composite/CompositeCOM.py and $RDBASE/rdkit/ML/Descriptors/DescriptorsCOM.py). The idea was to have descriptor calculation and contact to a predictive model from within Excel (or any other COM-capable app). This actually worked pretty well, and I like having the work done locally, but I was never happy with the way installation/deployment happened. If the deployment question could be addressed in a satisfactory way, I think COM provides a pretty straightforward (and well-documented) way of deploying chemistry into windows apps. Unfortunately it seems to be pretty windows specific, at least from the python point of view. -greg
