During the last decade, the development of optofluidic chips has become a large field of research. The integration of nano and microstructures with microfluidics layers allowed for the miniaturisation of a number of tools traditionally used in laboratories and optofluidic systems found their main applications in lab-on-a-chip and sensing platforms. At the same time, optical tweezers have become the tool of choice for the trapping and the manipulation of nano and micrometer-sized objects, that can be inert (dielectric or metallic particles) or biological (proteins, bacteria, viruses,...). In this context, nanostructures are the ideal candidate for tweezers miniaturisation, thanks to their ability to strongly confine light in small volumes and hence to generate intense gradient forces. This work reports on the fabrication of an optofluidic chip based on a two-dimensional photonic crystal cavity and on its use in particle and bacteria differentiation. The cavity has a hollow design that maximises the overlap between the confined field and the trapped objects and it supports Self-Induced Back-Action effects that allow for reducing the trapping power (down to few microwatts) and to simultaneously acquire information of the trapped specimen. The system is excited in an end-fire setup and the detection is carried out by recording the light intensity transmitted through the photonic crystal. The fabrication process, that is entirely carried out in the Institue of Physics and in the Center of MicroNanoTechnology cleanrooms, is first detailed. For the cavity design normally used, typical quality factors of 10000 were obtained. Moreover, SU8 mode adaptors were developed to increase the coupling with lensed fibers and a microfluidic membrane presenting two injection channels is proposed for rapid switch between liquids. The trapping and the differentiation of 250 and 500 nm polystyrene is then presented. The differentiation can be achieved qualitatively by direct observation of the transmitted intensity records and quantitatively by the use of histograms and of statistical moments. Finally, the trapping of seven species of living bacteria is shown and their Gram-type is determined by the analysis of the induced transmission increase. This ability to probe the cell wall of bacteria in a fast, label-free and non-destructive way paves the way for applications in biological and biomedical environments.