< 7^th term

Secondary structure assignment

Last update on the 19^th of December, 2019

In this task we use three algorithms to assign secondary structure: DSSP, Stride, SST — and compare them to the authentic PDB SS assignment.

Algorithms’ results and agreement

We used three web-servers that provide three SSA algorithms: DSSP, Stride, SST. One can download resulted assignments for DSSP, Stride and SST. We merged them with authentic PDB assignment into single file. Statistics of secondary structure elements per residue are shown in table 1.

All algorithms differentiate residues between turns and coils while unassigned residues in PDB were assigned to coils. DSSP introduces bends, DSSP and Stride — Isolated β-bridge residues, DSSP and SST — Right-handed π-helix. Right-handed 3₁₀-helix seems to be accurately assigned by the algorithms. The percentage of residues with identical assignments upon different markups is shown in table 2.

SST provides the most closest assignment to the PDB one. However, DSSP and Stride are in a greater agreement. What’s more, Stride is more identical to PDB and SST than DSSP to PDB and SST therefore suggesting SST to provide more biologically relevant assignment. That’s not confusing: SST is a more novel algorithm so it should be more accurate than the old DSSP and Stride to be accepted for publishing.

For in-detail view of discrepancies in assignments, we have drawn mesmerising figures of SS assignments (fig. 1A), their agreement with PDB (fig. 1C) and how much supported PDB assignment is (fig. 1B).

All the assignments generally agree at the extended cuts of α-helices and β-sheets but differ at their edges. Next we will discuss the agreement of assignments for selected SSE.

Case study

Here we will study 2 α-helices and 2 &beta:-strand and will discuss how well their ends are assigned. We will also cover some minor SSEs found in discussed elements. Visualization of hydrogen bonding will help us to decide which assignment is correct. Hydrogen bonds are constructed with dist command between donor and acceptor atoms at the distance up to 3.2Å.

α-helix 39-65

This is the common 4₁₃-helix (fig. 2).

Is 39 not in helix?

DSSP, Stride and SST propose such assignment. Indeed, residue 39 is not connected by hydrogen bond to the residue 43 (or any residue of the helix). What’s more, it is glycine, so that it is a convenient amino acid residue for looping.

Is 40 not in helix?

Stride and SST propose such assignment. Indeed, residue 40 is bonded to the sidechain of residue 44 thus not comprising α-helix.

Is 41-45 not in helix?

SST propose such assignment. However, these residues comprise regular α-helix except for residues 43-45 (no hydrogen bonding by C(O) acceptors). These assignment might be possible: residues 42 and 47 form S-S bond so the energy gain from covalent bonding probably outweighs the energy loss of leaving several possible hydrogen bonds loose. However, it still might be a deformed α-helix.

Is 46 not in helix?

DSSP and Stride propose that. That might be true: neither donor nor acceptor atoms of residue 46 form hydrogen bond with other donor and acceptor atoms of the backbone.

Is 65 not in helix?

DSSP, Stride and SST propose that. However, residue 65 comprises regular structure with residue 61.

Conclusion

Due to S-S bonding the beginning of the helix is not the regular one (lacking hydrogen bonds) but structurally resembles α-helix.

α-helix 80-105

This is the common 4₁₃-helix (fig. 3).

Is 80 not in helix?

DSSP, Stride and SST propose that. However, residue 80 forms a common hydrogen bond to residue 84 in an α-helical manner so proposition is wrong.

Is 81 not in helix?

SST proposes that. Again, residue 81 bonds to residue 85 in an α-helical manner so proposition is wrong.

Is 96-101 a π-helix?

DSSP proposes 97-101 to be a π-helix while SST proposes 96-99 to be that. It should be denoted that DSSP finds π-helices not shorter than 5 residues (SST-defined helix is of 4 residues). Indeed, residue 96 binds to 100 (α-helically) and 101 (π-helically), while residue 97 binds to residue 102 (π-helically). However, residues 98 and 99 binds to no residues, 100 and 101 bind to 104 and 105, respectively (α-helically). That is said, there is no long π-helix but only its seed. However, regular α-helical structure is deformed at that region so it is neither a continuous α-helix.

Is 105 not in helix?

DSSP and Stride propose that. However, 101 binds α-helically to 105.

Conclusion

Propositions concerning the ends of α-helix are wrong but there is a π-helix seed (96 to 102, 7 residues).

β-sheet

We will discuss a β-sheet comprised of the following β-strands: 110-121 (1), 124-129 (2), 28-33 (3) and 6-10 (4). The N-terminal part of the strand 1 interacts with strand 3, the C-terminal part — with strand 2. Strand 3 also interacts with strand 4. A certain turn between strands 1 and 2 is observed but not assigned in PDB.

Strands 1 and 2

Is 109 in strand 1?

SST proposes that. Indeed, it forms a hydrogen bond with residue 30 of the strand 3 in a regular manner therefore it must be included in the strand 1 (fig. 4).

Is 120 not in strand 1?

SST proposes that. However, it binds regularly to the residue 124 of the strand 2 so that’s not the case (fig. 5).

Is 121 not in strand 1?

SST proposes that. Indeed, 121 binds irregularly to the residue 124 (fig. 5).

Is 122-123 a turn?

DSSP and Stride propose that. Indeed, theese residues along with 120 and 124 form a canonical structure of a β-turn type II. Therefore 123 is not in the strand 2 as SST proposes (fig. 5).

Strand 3

Is 28 not in strand 3?

DSSP, Stride and SST propose that. Indeed, residue 28 does not bind to any other β-strand (fig. 4, 6).

Is 29 not in strand 3?

SST proposes that. However, residue 29 binds to residue 5 in β-strandish manner (residue 5 is adjucent to the strand 4) (fig. 6).

Is 34 in strand 3?

DSSP, Stride and SST propose that. Indeed, residue 34 binds regularly to residue 113 of strand 1 (fig. 5).

Is 35 in strand 3?

SST propose that. Actually, residue 35 binds to residue 115 of strand 1 but not in a regular manner thus it is not included in the strand 3 (fig. 5).

Strand 4

Are 1-4 in strand 4?

SST proposes that. However, they do not form hydrogen bonds in a regular manner with any β-strand residues (fig. 6).

Is 5 in strand 4?

DSSP and SST propose that. Indeed, residue 5 binds to residue 29 in a regular manner (fig. 6).

Conclusion

Strand ends are controversally assigned by PDB and the algorithms. Observations reveal that PDB assignment resembles native structure in some cases and algorithm assignment — in the other cases. SST deals the worst.

3₁₀-helix at 277-282

PDB annotates 277-282 as right-handed 3₁₀-helix (fig. 7). DSSP and Stride propose that 277 and 282 are not in helix. Actually, 277 forms hydrogen bond with 280 while 282 binds to 278 instead of 279 (279 do not form any hydrogen bond downstreams). SST proposes that the discussed structure is not a helix at all. But the 3₁₀-helical structure is clearly seen however posessing some deviations (residue 282 is supported with hydrogen bond from residue 278 instead of 279).

Conclusion

While SST provides the closest annotation to the PDB one, its results disagree more with the results of other algorithms and their annotations in selected cases.