Scientific Context of the Project

Reproducing the lost hard tissues (bone, tooth) using tissue-specific biomolecules has been a long-standing challenge. Rapidly expanding computational power and data analysis models offer tools for designing novel protein/peptide-based therapeutics by extracting the essential sequence-function relationship information, unprecedented just a few years ago. One of the key issues, however, is the difficulty of accessing powerful models that allow an analysis of the complex biological big data acquired from high-throughput experimentation. Community detection methods enable us to understand the structure of large and complex graphs representing different networks, including social networks, human cells, and protein-protein interactions. While current community detection algorithms improve the performance of the executions, very large networks consisting of a large amount of biological data, including biological units such as genes, proteins, individuals, or species, require further reduction in execution times to process complete graphs in reasonable times. Our project aims to design and develop efficient parallel community detection algorithms and implementations by accelerating target computations on multicore and GPU architectures. We will build both algorithmic and computational techniques utilizing parallel resources in target architectures. Hence, considering the biological data obtained from our collaborators, e.g., evolutionary selection of therapeutic peptides, virtual screening, and MD simulations, our design and implementation will process large graphs representing real biological networks. Our project will enable scientists working with large biological systems to discover the characteristics of networks efficiently. As part of this project, the development effort will open new directions to make our implementation available for various real-life problems.

Innovative Aspects of the Project

BIODETECT will develop computationally efficient algorithms for solving community detection for large graphs representing biological networks. We will utilize efficient executions with large data for designing novel protein/peptide-based therapeutics and tissue generation.

Research Environment and Infrastructure

PARS Research Group in Computer Engineering Department at IZTECH, works on parallel program development on multicore/GPU architectures. With graduate and undergraduate students, we focus on the optimization and performance analysis of parallel systems. While our local infrastructures provide compute resources, we have access to European supercomputers within the scope of the European High-Performance Computing Joint Undertaking (EuroHPC JU).

Required GRE Score

GRE Quantitative 157.00