CPU and parallelized GPU implementation
The folding of proteins is one of the biggest problems in every field of science related to Biochemistry. Before entering water solutes, proteins do not have a three-dimensional structure. After entering water solutes due to emerging forces between the solute and the protein as well as interactions inside the protein gives the linear protein a three-dimensional structure which is very important and fundamental to their functions inside a living organism. Non-bonded interactions are one of the most influential factors to protein folding. Which comprise of three major interaction: Vander Waals interactions, electrostatic interactions and hydrogen bonds. Our goal in this project is to compute all of the interactions of these kinds inside a PDB file and parallelize these computations using GPU programming. For this purpose, we use a parallel computing platform called CUDA. We use python Numba package to parallelize computations. In the end, we compare the computation time of the CPU program and GPU program. This comparison shows that increasing number of computation, CPU time exponentially increases and GPU time linearly increases and GPU Time is lower than CPU time.
Seyed Ali Mirmostafa: M.Sc. student at Bioinformatics laboratory of Tarbiat modares University (S.mirmostafa@modares.ac.ir)
Maryam Rafieipour: M.Sc. student at Bioinformatics laboratory of Tarbiat modares University (r.maryam@modares.ac.ir)
Niloufar Seyed Majidi: M.Sc. student at Bioinformatics laboratory of Tarbiat modares University (niloufarseyedmajidi@modares.ac.ir)
Rayeheh Vafaiee: Ph.D. student at Bioinformatics laboratory of Tarbiat modares University (rayeheh_vafaee@modares.ac.ir)