Project Critique (Revised)
Advanced Research Topics in Computer Science (CSC 181)
By Parag Jain
The project for this course stands in good position as of now. Everybody in class seems to be interested in the project and is putting great effort. The goal of the project, viz. to compare several molecular surface computation algorithms, is very reasonable and achievable. The project is absolutely worth pursuing and might be quite useful keeping in mind the ongoing research work in the field of molecular biology. I like working on the project a lot - I have been working on the project for more than two months now but due to the vastness and intricacy involved in the project I still have a lot to work on and study within the project. I always feel motivated to work on the project and learn more each passing day.
I believe I have a comprehensive project webpage that contains all the information pertaining to my efforts on the project. The webpage has links to the algorithms that I have used to compute the molecular surfaces and delaunay triangulations, and also the results that I have obtained from using these algorithms. On my project webpage there are links to those websites that I have studied and gathered information from; in conjunction to that there is a bibliography section on the webpage that is a list of the papers and journal articles that I have referred to while working on the project. I have tried my level best to keep my course webpage up to date and put all my day to day findings and also all my project submissions and assignments on the webpage. I believe that by the end of this semester I shall put more information on my project webpage so that anybody who visits the webpage knows what the project was about and what the final results were.
However, I am not completely satisfied with my webpage right now because there is very little rather no text on the webpage - the webpage contains only links to other webpages. Even though the links are self-explanatory, I wish to put more text on the webpage to explain the project process. The algorithms and the results that have been presented on my webpage are arranged in the order that I worked on them in the real world. I worked on a divide and conquer algorithm that computes the Delaunay Triangulations for a molecule. The divide-and-conquer algorithm subdivides the area into two partial areas, computes recursively the Delaunay triangulation of the partial areas and merges finally both triangulations. After that I worked on DMS, an application that calculates the solvent-excluded surface of a molecule. The molecular surface resembles the van der Waals surface of a molecule, except that crevices between atoms are smoothed over and interstices too small to accommodate the probe are eliminated. The surface includes cavities in the interior of the molecule, even if they are not accessible to a solvent molecule coming from the outside. I later worked on another molecular surface computation algorithm called MASKER that is an improved solvent excluded molecular surface area estimation using Boolean masks. This algorithm also calculates the solvent-excluded surface of a molecule. However, this algorithm is based on the boolean mask approach that was proposed by LeGrand and Merz (Le Grand and Merz 1993). This algorithm takes the method of Le Grand and Merz one step further by using masks to calculate the solvent-excluded surface. As of today the plot of the times that both DMS and MASKER took to compute the surface area versus the number of atoms in the molecule fits well with the graph of a second degree polynomial.
Coming back on the topic of the webpage, I strongly believe that we all need to work on the common course webpage because it is not to informative and descriptive as of now. We have at certain occasions tried to put some effort to make the common course webpage look better but work still needs to be done on the webpage. We need to make it more organized such that it becomes very clear to anybody who visits the webpage as to what the project was about. Apart from that, I strongly think that we need to come up with a common scale on which to plot the graphs. I am using an application called ‘DataFit’ to plot graphs for the results that I obtained from DMS and MASKER. I am aware of the fact two other group members are using the same application. We should decide that we all would use the same application so that the comparison becomes easy more accurate.
Also, we should discuss if there are some specifications, that we all have agreed upon, which are incompatible with any group member’s algorithm, and if there are any then we should modify the specifications so that everybody can conform to them. For example, Vince was having problems in getting his algorithm to run for some specific pdb files; such pdb files should not be test run by anybody else as well. The idea is to have coherence in the result sets of everybody in the class. Another aspect that we should take care of is the timing of the algorithms, i.e. we should all try and use the same timing algorithm to make the results more comparable.
I believe that the overall structure of the course is good and there is less that we could do to improve it. There is enough collaboration between the group members and there is some consistency (if not a lot) in the group members’ results. All in all, the project is in good standing and I can not think of any other demand from the project apart from the ones that I mentioned earlier. I think that we should be able to succeed in what we are trying to do in this project if we keep working at the same pace as we are now. The effort that has been put into the project is appreciable and I am sure that our work will be of use for somebody who wishes to do further research in this field.