 Dear students, now we'll move forward in our review of structure prediction strategies. You already know that there are three common strategies that are employed towards predicting protein structures by using their sequences. The first one was homology modeling, the second one was fold recognition or threading and lastly ab initio methods. So in this module we'll be reviewing fold recognition. So in this slide you can see that if we had the primary sequence of proteins that was obtained from either admin degradation or mass spectrometry then we could move towards homology modeling. However, if we had no homologs with 3D structures then we had to move to a different strategy. This strategy is called fold assignment. So once again if you cannot find complete protein structures having similar sequences to your sequence then you have to move to fold recognition. So in fold recognition you started by looking at folds from the library. So these folds were then investigated individually for possibility of being formed by the amino acid sequence. A protein may have multiple folds of course. So here is your amino acid sequence, here is the scoring function and all of that is put into this threading algorithm and you may now know that the most common tool for threading is the eye tesser. You should go and use it sometimes and then the threading software it outputs the predicted structure. Now let's take a look at a step by step flow chart. So you started by looking at the sequence of these proteins. You found the homologs from the sequence databases by blasting them. These were called the templates and then you went for finding the structures of each one of these templates from the PDB and then these PDB structures were searched in the fold database such as SCOP. So for each PDB file you will be having multiple folds, fold 1, fold 2, fold 3. So you will have a list of folds that each structure has. Next you compare the predicted structure and experimental homologs. So if the structure that is predicted is plausible and functional then you select it and you try to create a tertiary structure by creating combinations of the folds and therefore your structure is made. So in conclusion for the fold recognition you may have a low identity and a low sequence alignment score. So in such cases threading is very useful. So threading relies on the fold databases and you have to find different folds from the overall protein structure and then investigate each fold individually to form that fold from the sequence that you have. Also the fold recognition paradigm is extremely useful in cases for novel protein structure identification. So if your protein structure is not there in the PDB then you can use threading towards predicting novel protein structures as well.