There are tens of thousands of 3D structure in PDB,and a portion of them are fragments but not full-length due to some flexible motif will adopt disordered conformation.The question is that, if we build a homologous model of a full-length protein based on its known fragment structure ,and then minimize the energy of the resultant model (via moe or other software), then how confident are we of the putative 3D structure of the full-length structure?
Well, if I understood you right, some parts of the full length protein won't be in the scope of the template you will be using for the modeling. Since the classical homology modeling approach is relying on a single template structure, this is very problematic. It is very difficult to predict the quality of the resulting "missing" parts. They could be just alright, but they could also be totally wrong. A normal force field will not be able (and wasn't designed) to generate a "correct" fold during a minimization "from scratch".
In this case, I would suggest you to use a more advanced methodology like threading. In these approaches, the combination of several templates and the use of fragment libraries significantly enhance the model quality. In any case, the usage of protein structure validation tools (Prosa, Procheck, QMEAN) would be a good idea to get an impression of the model confidence.
Hi Lin, I agree with Felix.
Because with unordered residues it can be a problem to verify the actual structure they acquire and by using just one template it can make things more difficult for a good structure prediction. A good way out can be to use multiple threads/fragments for your proteins and then you can minimize the structure using MD simulations to see if the predicted structure goes well or not.
Hi I agree with Felix, there is no guarantee that built 3D structure would be absolute prediction, I therefore suggest to perform PROSWEB analysis to see closeness of the structure with crystal structure. Then perform MD simulations which will give considerable answers to your questions
You can decipher the structure. Its called THREADING. finding the structure of a protein based on the fragments of the template knows. If the homology is more than 80% its trustable to use homology modeling. If its less than 60% threading is a good option. i-Tasser is a threading webserver
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology