Suppose I have a protein chain. And I have a hydrogen bond detection algorithm that can detect hydrogen bonds in a protein chain. Say, the name of the algorithm is HBDA algorithm.

How can I detect secondary structures in a chain using the above? I.e., what rules can determine the secondary structures in that chain?

Google says that:

  • Alpha helices typically show a repetitive i to i+4 pattern in the C-alpha backbone
  • In parallel Beta sheets, bonds generally form between residues i and i+2, while in antiparallel sheets, bonds form between residues i and i-1 (or similar close segments)

Is that correct?

So, what will be the rule for coils or loops?


1 Answer 1


Firstly, of course, I should point out that there is software that already does this DSSP and Stride, among others (see a partial list here : https://2struccompare.cryst.bbk.ac.uk/methods.php)

However, I have no particular problem with reinventing the wheel on this - as I have once looked at doing to for SSE assignment! - so some issues to note are:

  1. Hydrogen bonds by themselves are not enough to assign SSEs
  2. The backbone angles (phi, psi) are also required
  3. Loops (in the sense of short defined loops) have specific angles but coil or large loops are just the parts that are not helix or strand

It could be helpful to look at the original papers for DSSP and Stride, as they describe their methods. There is also DSSPcont (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC169032/) although their website does not resolve anymore.

Broadly, an algorithm for HBDA would be :

  1. Determine the backbone angles and hydrogen bonds -> (A, H)
  2. Look for continuous stretches of helix-like and strand-like angles in A -> S
  3. Check the bonds in H between or among the elements of S
  4. Classify the elements found into types, with remaining unassigned residues as coil

So step 3) is slightly redundant for helices as the backbone angles should totally classify stretches of more than 5 (roughly?) residues. You may want to distinguish alpha, 310 and pi helices here - although there are few pi helices.

For strands and sheets, though, the hydrogen bonds are important for determining whether a stretch of backbone is part of a sheet (and therefore a strand) or just extended coil. Finally, for tight loops (turns) the hydrogen bonding is just detail as you only get 'real' turns between strands in a sheet.

  • $\begingroup$ Putting this as a comment, so that the answer is standalone, but I did see your other question bioinformatics.stackexchange.com/questions/22434/… so I know you know about DSSP/Stride :) $\endgroup$
    – gilleain
    Commented Apr 24 at 12:43
  • $\begingroup$ This question was posted before that question. I have done some studying in the meantime. But, please try to reply to that question, too. $\endgroup$
    – user366312
    Commented Apr 24 at 13:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.