Progress Report - 20 Jul 2007

1. What were my goals for this week?

• Implement k-means clustering
• Find other potential clustering algorithms, read and implement
• Try some of them out on real data!
• Discuss possible better critical point selections with Aaron

2. What did I accomplish this week?

• Wrote several versions of k-means algorithm
• Got back into the swing of C++!
• Heard Rachel's defense
• Tested k-means on (very limited) data
• Fought with Aaron's MoleculeLoader for a while

3. Why are numbers 1 and 2 different?

Got hung up on C++ again, but I think I'm getting back into things. The file reading is going to be obnoxious - I can't get mmdb to build properly with the Molecule project and I think just leaving for now instead of wrestling with the ideal of testing on real data is going to be better in the long run. Hopefully either Aaron will be in on Monday or I'll have figured out how to get mmdb_manager.h to include properly when Molecule builds.

Didn't actually get around to implementing other clustering algorithms in favor of using k-means as an exercise to get back into the proper mindset for implementing in C++.

4. What are my goals for next week?

• Get moleculeLoader working with my project
• Write a routine to spit out the clusters into a PDB for viewing in PyMol or somesuch
• Test algorithms with position data to see if they work
• Look at possible ways to extend this to take motions into account?

5. How does this fit in my bigger picture?

Extensions of this project will help with the motion representation - if we can cluster by motion AND position, it should be easier to come up with a good representation of the global motion of the cluster.

6. What did I read this week?

• Kundu, Sibsankar, Dan C. Sorenson and George N. Phillips, Jr, Automatic Domain Decomposition of Proteins by a Gaussian Network Model.
• Meyers, Scott, Effective C++ (Third Edition).