Student Posters Repository

In the last few decades due to new high-throughput techniques, such as mass spectrometry, we gained a large amount of data about Protein-Protein Interactions and Domain-Domain Interactions. This opened the opportunity for computational methods to predict protein complexes. Here we propose a simulation based protein complex prediction tool, which uses Gillespie's Multiparticle algorithm to calculate how the composition and abundance of protein complexes changes proteome-wide. The approach combines the classic Gillespie's algorithm with diffusion by splitting the simulation space into several thousands of sub-volumes which are small enough that we can assume that the reactants are able to find each other. In given times the molecules diffuse between the SVs and participate in binding and unbinding reactions. An earlier prototype could run these simulation with a greatly reduced number of proteins and only considered a two dimensional space. In our new simulation tool I implemented a three dimensional simulation space and improved the performance. To achieve this, we are using OpenMP for CPU level parallelization and planning to use MPI as well as GPUs to speed up spatial simulations for a further increased performance.