I am interested in experimenting with folding simulations and algorithms for arbitrary sequences. I'm wondering if there is an easy way to convert an amino acid sequence into a PDB file for further simulation. To be clear I only want the primary protein structure.

If possible, I'd like to be able to characterize the bonds as well so that I can treat the molecule as a rigid body with rotational joints. Does anyone know something that can do this?

  • $\begingroup$ You may be confused about the terms. If you only want the primary structure you don't need a PDB file. There is no easy way to "convert" sequence into 3d model (PDB file), but the opposite is true - you can get sequence from a PDB file. $\endgroup$
    – marcin
    Sep 18 '17 at 9:00
  • $\begingroup$ @marcin - I believe he does need a PDB file (or cif, or xyz, or whatever) as it would be an unfolded 3D model, with coordinates for all the atoms, but without a realistic geometry. As in fold.it, basically $\endgroup$
    – gilleain
    Sep 18 '17 at 12:12
  • $\begingroup$ yes, exactly @gilleain -- I want the 3D geometry of the unfolded model. It seems to me this is a relatively easy task since the geometry of the amino acids is known ... you just need to stack them together and preserve the peptide bond length. I'm just wondering if there is a library for this or do I need to write one. $\endgroup$
    – Salik Syed
    Sep 22 '17 at 22:20
  • $\begingroup$ @SalikSyed No, from what I know of there is no library that offers this description of a protein. I say this with a significant amount of confidence, as I myself just went through a similar search, and which resulted in me manually defining my own library, however, my library doesn't initially offer coordinate data, but instead, takes on a more topological perspective. It's written in Java, should you be interested. $\endgroup$
    – user22020
    Oct 7 '17 at 21:05
  • $\begingroup$ Also, I would like to mention that PDB files specifically represent folded proteins, which includes 3-D coordinate data. Having this 2-D coordinate data is pretty much useless for the scope of interest and nature of research for most all scientists within this field. $\endgroup$
    – user22020
    Oct 7 '17 at 21:07

So to be clear, it sounds like you want the 'coil' or unfolded structure of a protein based on the sequence?

There are plenty of programs out there to do homology modelling, which is taking a sequence of unknown structure and modelling it onto one with a known structure. On the other hand, there are many libraries for analyzing existing structures. What you want is somehow in between these two.

I'll add some links here as I find them:

I'm sure there are more, but it really depends on what programming language you want to use, or if you are implementing stuff yourself versus running some standalone software.

  • $\begingroup$ Thanks. Yes, I explicitly do not want the folded structure or to predict the folded structure via analysis or analogies to existing structures. I just want to convert the amino acid sequence into an unfolded 3D configuration of the protein where bond lengths are close to their average lengths. I want to then explore the conformation space solely based on perturbations to torsion angles (sample the energy for various configurations of the protein backbone) starting from the unfolded state. $\endgroup$
    – Salik Syed
    Sep 22 '17 at 22:07
  • $\begingroup$ rFold seems to be exactly what I am looking for, though I think I need to specify the relationships, native bond lengths and dihedral angles for each amino-acid manually $\endgroup$
    – Salik Syed
    Sep 22 '17 at 22:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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