Chou-Fasman Method Overlapping Regions Problem - Secondary Structure Prediction

Question

According to the Chou-Fasman method, when there is an overlapping region on the amino acid sequence, we compare the scores of alpha-helix and beta-sheet and chose the highest one. Let's say I have this amino acid sequence: ABC.

As a result of computations, AB is identified as alpha-helix, and BC is identified as beta-sheet. This means, I have an overlapping region, which is B in this case. The algorithm says that we need to compare the scores of the overlapping region, and select the secondary structure with the higher score, let's say alpha-helix in this case.

Would the resulting secondary structure be $\alpha$$\alpha$$\beta$ or $\alpha$$\alpha$_?

In other words, do we cancel out the secondary structure identifications when we encounter overlapping regions?

Thanks in advance.

Answer 1

Sure, an isolated residue has a ψ and φ angles, so you could technically have an isolated residue in a given SS conformation, but that does not mean much if its neighbours' backbones aren't in that same conformation. Also, sure, in Rosetta design algorithms residues are thrown into ABEGO bins based on ψ and φ angles alone and as a result one can get isolated α-helix bin residues (in what is a loop): but that's a design algorithm that will want to correct this.

In your particular example, IIRC that algorithm uses a window of 4 or 6 residues for SS prediction. So it will surely be (αα)αα_. But I am pretty certain you cannot have a ααααββββ switch without a loop or turn in nature —I might even hazard a guess that for design it is not Koga-compliant even if not specifically addressed. And with the Chou-Fasman algorithm you'd need to have the windows spanning the switch score above the threshold in the correct way. I am sure there's a combination that allows it as after all we're talking of an ancient 70s prediction method based on probabilities alone...