Online classification tools

I have just set up some online classification tools based on the WEKA machine learning library.
More information can be found at: http://edcannon.hobby-site.com/SimpleUpload.jsp .

Basically people can upload one file upto 5mb in size and then use either: SVM, NBayes, Logistic, k-NN (k=1), RandomForest or the PART rule based learner.

The results are based on the default parameters in WEKA and using 5-fold cross-validation.

I hope some people might make use of this facility!

Webservices

Initially I thought today was not going to be such a productive day. However, I have now managed to set up an online web server http://edcannon.hobby-site.com/CDKDescriptorCalculation.html) that can calculate descriptors from the Chemistry Development Kit given a smile string as input. The descriptors that have been implemented at present are:

• BCUT
• Fingerprint
• Apol
• Number of aromatic atoms
• Autocorrelation descriptor with polarisability
• Number of hydrogen bond acceptors
• Number of hydrogen bond donors
• Lipinski's rule of 5
• Total polar surface area
• XlogP

More descriptors will be made available later. I also intend to put an online service for machine learning algorithms in using the Open Source WEKA Java library.

Off to bed.

Nearly There

Just the bibliography and the appendices to sort out....thank God its taken a while.

I now need to add a few more programs and scripts that I wrote during my PhD to my wiki, then add a few tags to make use of the semantic nature of the wiki.  Now that I have a bit of spare time, I am starting to look at RDF, RDF Schema, SPARQL and OWL, I am also investing more time in Taverna.

==Chemoinformatics paper of the month?....== 

TBA next post!