i After two weeks I finally began working on and testing algorithms that are directly related to my project. The purpose of our Ligand Binding project was to compute metrics to determine binding sites on a protein sample for a ligand sample. To do this, we made use of the Uniform Obstacle Based Probabilistic RoadMap (UOBPRM). This sampler evenly distributes the generated samples around an obstacle to avoid clustering. Though not perfect, it helped us get a good idea of what we wanted. In this project the protein was the obstacle and the ligand the robot, and our goal was to get the ligand to a good binding spot on the protein.

During my third week, I focused on testing the sampler I would be using for my project. I ran experiments changing different parameters for the UOBPRM sampler, then I generated data files (runtime, collisions, number of edges, nodes) using a perl script. After this I used the data files to plot graphs using gnuscript and GhostView.

I attended our Motion planning meeting where I listened to a sample presentation a motion planning conference. Opinions were given at the end by different persons to make the presentation better, and questions related to the material presented were answered. The presentation was helpful to me because I got to take a few points on making a good presentation.

Next, I met with my teammates on the project to make clarifications on some things I did not understand about the sampler, and also to get more information on the project and our next step towards it. I started reading on a Ligand Binding paper to understand more about how motion planning applies to Ligand Binding. This paper helped my transition from robotics to ligand binding with motion planning.

I also attended some C++ lectures given by a PhD student. We went through C++ classes and basic function properties, which later helped me on the assignment from the lectures. Then I ran more tests with UOBPRM and OBPRM (Obstacle based Probabilistic RoadMap). The tests were on different 3D environments to compare how fast they run and their data outputs. The OBPRM differs from the UOBPRM because the former only serves to generate samples close to an obstacle but often clusters the samples around some region of the obstacle. After running tests on the sampler, I figured why it was essential to have the samples generated uniformly for our project and some other nice perks of the UOBPRM sampler.