When I tried to analyse the structural diversity of a peptide and its mutants, I found it is difficult to connect the probability of hydrogen bond formation of a residue with its propensity to acquire one kind of secondary structure. For example, residue A(i) exhibited considerable propensity to obtain α-helix, but the probability of backbone hydrogen bond formation between residue i and i + 4 was low. How to explain this phenomenon? STRIDE algorithm was applied to assign secondary structure of protein.
Bartosz makes excellent points. I would also add that STRIDE uses a complex series of metrics, including numerous factors related to hydrogen bond geometry, planarity, energy function, etc. as well as dihedral potentials, to determine the type of secondary structure. It is more complex than the DSSP algorithm, which is just based on hydrogen bonding, though the two agree frequently.
Lulu: no algorithm is perfect.
Have in mind, that basically all amino-acids (apart from P) can form i ... i+4 hydrogen bonds and even proline is commonly found in stable helices, though it introduces a kink. For example GPCRs, consisting of 7 transmembrane helices, have all 20 natural amino acids present in their helices. So a simple property like the propensity of amino acids to form alpha-helical structure does not tell the whole story.
The actual problem of solving the secondary structure of a protein is not yet completely solved. We, when studying GPCRs and trying to figure out where do the transmembrane helices start and end, use 3-5 different secondar structure prediction servers, including always PsiPred, Porter and APSSP2 and sometimes adding Jpred and SSPro and form a consensus prediction out of those. Even in such case we are sometimes off by 1-3 residues for the prediction.
Bartosz makes excellent points. I would also add that STRIDE uses a complex series of metrics, including numerous factors related to hydrogen bond geometry, planarity, energy function, etc. as well as dihedral potentials, to determine the type of secondary structure. It is more complex than the DSSP algorithm, which is just based on hydrogen bonding, though the two agree frequently.
- Badges
- Science topic