The main purpose of structural genomics is the identification, in large scale, of possible targets for the action of bioactive molecules such as inhibitors, drugs and vaccines. Due to experimental difficulties, only a small number of identified sequences in genomes have their 3D structure in the protein structure databank (PDB). Bioinformatics search of similar sequences and homology could help to construct models for the new sequences. Further molecular dynamics simulations could help to test bioactive molecules and intermolecular interactions. We have developed a computational pipeline (MHOLINE) to analyse in large scale the sequences from genomes, adding structural information to the usual bioinformatics annotation, and when possible to propose structural models for the new sequences. Further refinements are achieved by using molecular dynamics simulations and analysis of interactions at atomic level. Examples are presented concerning targets of biotechnological and biomedical relevance.