The future of computational science is facing serious challenges: until recently, the computing power of processors was quickly increasing. However, physical limitations in the miniaturisation of computer chips halted this steady source of performance gains. Paradoxically, the computational cost of problems awaiting to be solved never stops to augment. Breaking these computational barriers is therefore a central concern that emphasises the crucial need for parallel and distributed computing in the future of computational science. In this thesis, we take part to this development. First, we study the potential of using graphical processing units to accelerate the Simplex algorithm. Then, we investigate if the use of highly-parallel computational fluid dynamics could broaden our understanding of the mechanism of the hearing organ. Finally, we tackle the computational challenges raised by the study of Darwin's theory of evolution that stem from complex statistical models and the ever growing amount of molecular data.