Motivation: Structure-based Computational Protein design (CPD) plays a critical role in advancing the field of protein engineering. CPD seeks to identify minimum energy amino acid sequences that allow proteins to fold into a given three-dimensional structure and ultimately perform a targeted function. CPD is essentially the inverse problem associated to protein structure prediction. The combinatorics on protein sequences and the number of degrees of freedom of protein structures makes CPD a challenging task from a computational point of view. Results: We propose a data-driven computational protein design method that exploits information on amino-acid local structural environments of protein known structures to guide sequence space search, while sampling side-chain and backbone conformations to accommodate the mutations. Our method, Shades (Structural homology algorithm for protein design), is based on the construction of customized libraries of possibly non contiguous amino acid residue motifs that are in contact in 3D protein known structures (called, In-contact residue Tertiary Motifs or ITEMs) and their use for the design of full protein sequences. We have tested Shades on a public benchmark of 40 proteins selected from different protein families. When excluding proteins homologous to those from the benchmark, Shades achieved a protein sequence recovery of 30% and a protein sequence similarity of 46% on average. When homologous structures were added, the sequence recovery rate increased to 93%. Availability: Shades source code is available at https://bitbucket.org/satsumaimo/shades. The source code contains a patch for Rosetta version 3.8 and a tool for protein structure database and candidate ITEM library creation.