DMP Research : Summer 2004

  Your Logo Here 


Weekly Journals

Week 8 ( July 20th ~ July 26th):

 

In the last couple of weeks, I have completed the contact region detection for frames. This week I will extend it to the trajectory level. My idea is that since each trajectory has a sequence of frames where a protein is in a different conformation, then since I am now able to achieve contact region detection for each frame correctly, I will get a stream of contact information.

An important benefit of using contact streams is that by generating the contact stream for a plethora of protein trajectories, and applying pattern recognition techniques, I can isolate patterns for different types of trajectories, such as those where the residue chain folds into a loop or forms three bends. Another benefit of the contact stream is that I can easily tell the critical frames apart from ordinary frames in a trajectory. The first benefit of the contact stream is already enough to serve my project here. Because of the noise patterns from the contact stream, I am considering to implement a noise-filter in the upcoming week.

 

 

Week 9 ( June 27th ~ August 2nd):

 

            In this penultimate week, I have determined that noise occurs unpredictably in the contact region detection among frames. However one thing that makes this mission still possible is that the duration of these noises is relatively short. Thus, a linear filter will be sufficient to filter them out, and no secondary filter is necessary. So after the filter is add, now my program can take in any collection of trajectories, categorize them based on the contact information extracted from each conformation in the trajectory.  It then is able to tell what kind of trajectory it is such as those forming loops, asymmetric/symmetric one bend, two bend, etc. Furthermore, the program can categorize all the input trajectories into different types as well.

           All the user needs to do is to provide a threshold, including a cutoff and window size. This threshold is computable from the interatomic distance, and thus the whole critical event detection and trajectory analysis achieves automation.

 

 

 

Week 10August 3rd ~ August 9th):

 

How time flies! My summer research project has reached its concluding week. Looking back, I have gained new knowledge in both bioinorganic and computational techniques in the past nine weeks. I am happy that I accomplished my tasks. Given any complex protein trajectories, I can perform critical event detection by the contact stream method mentioned before, and then "cut" the complex trajectory into simpler sub trajectories. Next, I generate and analyze the contact stream for each sub trajectory, and finally being able to determine the characteristic of each trajectory and further categorize them. This project can indeed expand and accomplish much more. Moreover, I really wish I could have more time and be able to continue with the work. DMP this summer is again, a great experience for me. I not only learned a lot and applied what I learned to complete the project, but most importantly, I demonstrated my ability to pick up new ideas and sharpened my ability to learn. 

 

 

Previous Page

 


Lin Kuang

Contact Info: klin7. AT. berkeley.edu




Home

Weekly Journals P1

Weekly Journals P2

Weekly Journals P3

Final Project Report