This section presents the seminar about homology modelling as a series of short videos. In this short version the focus is on the act of homology modelling for drug design. The (many) underlying Force Fields will be discussed separately.

Figure 14. Before building a model we should ask two questions: Why are we building a model? And, suppose we've got the model, will it be good enough to answer our scientific question(s)? After all, modelling is a bit like gambling...

Figure 15. This is the most crucial plot for anybody who actually wants to build a model by homology. If your alignment falls above the curve you can build a model. For drug design, though, you want at least the active site area to lie far above the curve.

Figure 16. We will use an 8 step process to explain modelling. Not that it really IS an 8 step process, but it explains nicely this way. The first step is the detection of the template from which to build the model.

Figure 17. Software like BLAST produces an alignment, but one normally can obtain a better alignment with dedicated alignment software augmented with common sense.

Figure 18. Generation of the model backbone can be highly complicated, but when drug design is your goal, you better keep this step simple.

Figure 19. Building loops is an art, not a science, albeit that the structure of very short loops often can be predicted reasonably well.

Figure 20. There are two main solutions for placing the side chains of the residues that are different between model and template: energy based and rotamer library based. We will discuss how that can be done using rotamer libraries. Later during this course you should be able to work out all by yourself how this can be done using energy calculations.

Figure 21. Molecular dynamics is often used to improve models. Unfortunately most MD programs make models worse rather than better.

Figure 22. No matter how good the software and how careful the modeller, at the end the model will contain errors. Errors are less of a drama when you know about them, and that is where structure validations comes in.

Figure 23. If the validation process detects errors that can be corrected by doing certain steps of the modelling different/better, then obviously that should be tried.

Figure 24. Some closing remarks.