Localized surface plasmons (LSP) can be excited in metal nanoparticles (NP) by UV, visible or NIR light and are described as coherent oscillation of conduction electrons. Taking advantage of the tunable optical properties of NPs, we propose the realization of a plasmonic structure, based on the LSP interaction of NP with an embedding matrix of amorphous silicon. This study is directed to define the characteristics of NP and substrate necessary to the development of a LSP proteomics sensor that, once provided immobilized antibodies on its surface, will screen the concentration of selected antigens through the determination of LSPR spectra and peaks of light absorption. Metals of interest for NP composition are: Aluminium and Gold. Recent advances in nanoparticle production techniques allow almost full control over shapes and size, permitting full control over their optical and plasmonic properties and, above all, over their responsive spectra. Analytical solution is only possible for simple NP geometries, therefore our analysis, is realized recurring to computer simulation using the Discrete Dipole Approximation method (DDA). In this work we use the free software DDSCAT to study the optical properties of metal nanoparticles embedded in an amorphous silicon matrix, as a function of size, shape, aspect-ratio and metal type. Experimental measurements realized with arrays of metal nanoparticles are compared with the simulations.