 Dear students, in this module we will be reviewing the ab initio methods that we have studied towards structure prediction. You may remember by now that there are three general strategies, homology modeling, whole recognition or shredding and ab initio methods. So in this module we will just review the ab initio strategy. You know that if you have the protein sequence with you and if you find very nice homologues from the PDB then you can do homology modeling. But if you could not find such homologues then you can go towards fold assignment or fold recognition. But there can be a case where such folds are also not available. So you are left with no way to use homology modeling or fold recognition. So in such a case you go for ab initio modeling. So in ab initio modeling there are two steps and here I will discuss the first step. You start with a rough initial model so you estimate your protein to have an alpha helix and a beta sheet followed by another alpha helix. So this is your rough model. This is also known as the initial model. So next you define an energy function that counts bonded and non-bonded atoms within the structure and the overall energy is just the sum of these two factors. So the overall energy has to be minimized in order for you to obtain the optimal structure. This problem needs to be solved computationally and you have to find the global minimum. You may remember that the global minimum is not an easy thing to find. You may find a local minimum instead of a global minimum and you may think that this in fact was a global minimum. So in order to avoid such situations you need to have a very nice energy function. Okay towards the second step of ab initio. So in ab initio modeling once you have predicted your protein structure and optimized its energy this will help you to form an accurate model and this will obviously include the energy and the forces. So next you place it in a force field in order for you to study the dynamics of the folding process itself. Once you set this simulation into motion the folding process will continue for a long time and eventually after the simulation achieves a low energy model. This will be the native structure which will emerge from this process. So this was all for ab initio modeling and please remember that ab initio modeling is useful in those cases where homology modeling and threading fails. However the downside is that it is computationally extremely expensive and you can only predict the structures of very small proteins by using ab initio modeling. So you may want to predict the structures for small sequences and then create threading later. So by combining the small structures that are predicted from ab initio modeling you can try to see how their combination can produce the overall protein structure.