The construction of models of protein structures before experimental data is available is an important part of drug discovery and protein engineering. The key steps are to generate an appropriate sequence alignment and then convert an existing structural template. Sequence alignment techniques and sequence threading methods both have their limitations since so much is dependent on spatial arrangement and side chain packing. This talk will approach all of these steps in new ways. Sequence alignment becomes transformed when we look at aligning on conserved core regions and allow for hypervariability in loop composition and size. The use of a simplified atomic potential, rotamer/conformer libraries and genetic algorithms allows for the rapid construction of complete models for evaluation. The details of these methods will be discussed. Lastly the development of collaborative tools in informatics to support genomics, proteomics and drug discovery will be discussed.