Limits...
An Algorithm for Protein Helix Assignment Using Helix Geometry.

Cao C, Xu S, Wang L - PLoS ONE (2015)

Bottom Line: The first step searches for a series of bona fide helical curves each one best fits the coordinates of four successive backbone Cα atoms.The second step uses the best fit helical curves as input to make helix assignment.The structural uniformity should be useful for protein structure classification and prediction while the accurate assignment of a helix to a particular type underlies structure-function relationship in proteins.

View Article: PubMed Central - PubMed

Affiliation: The College of Computer Science and Technology, Jilin University, Changchun, Jilin, China.

ABSTRACT
Helices are one of the most common and were among the earliest recognized secondary structure elements in proteins. The assignment of helices in a protein underlies the analysis of its structure and function. Though the mathematical expression for a helical curve is simple, no previous assignment programs have used a genuine helical curve as a model for helix assignment. In this paper we present a two-step assignment algorithm. The first step searches for a series of bona fide helical curves each one best fits the coordinates of four successive backbone Cα atoms. The second step uses the best fit helical curves as input to make helix assignment. The application to the protein structures in the PDB (protein data bank) proves that the algorithm is able to assign accurately not only regular α-helix but also 310 and π helices as well as their left-handed versions. One salient feature of the algorithm is that the assigned helices are structurally more uniform than those by the previous programs. The structural uniformity should be useful for protein structure classification and prediction while the accurate assignment of a helix to a particular type underlies structure-function relationship in proteins.

No MeSH data available.


The illustration of the differences in assignment by our algorithm and dssp.In (a) dssp assigns the entire segment (51–65, excluding P50) in a protein (pdbid 3OY9) as an α-helix. Our algorithm divides it into two helices: 310-helix (51–52, red) and α-helix (53–63, green, purple, yellow). The α-helix stops at N63 since the Cα RMSD δ values for residue 64 and 65 are respectively 0.541, 0.431, none of them less than dmax = 0.3. In contrast, the dssp assigned helix extends to residue 65. However, as shown in the left figure, the Cα coordinates of both residue 64 and 65 deviate clearly from a helical curve. In (b) a segment of residues 153–172 in a protein (pdbid 1MHY) is assigned as a single α-helix by dssp while our algorithm divides it into four helices: 310-helix(154–156, yellow)–α-helix(157–163, green)–310-helix(164–166, purple)–α-helix(167–171, green). However, a careful examination of the hydrogen bond energies for these residues in fact suggests that they could also be assigned to a 310-helix even by the dssp standard (S2 Fig).
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4488512&req=5

pone.0129674.g005: The illustration of the differences in assignment by our algorithm and dssp.In (a) dssp assigns the entire segment (51–65, excluding P50) in a protein (pdbid 3OY9) as an α-helix. Our algorithm divides it into two helices: 310-helix (51–52, red) and α-helix (53–63, green, purple, yellow). The α-helix stops at N63 since the Cα RMSD δ values for residue 64 and 65 are respectively 0.541, 0.431, none of them less than dmax = 0.3. In contrast, the dssp assigned helix extends to residue 65. However, as shown in the left figure, the Cα coordinates of both residue 64 and 65 deviate clearly from a helical curve. In (b) a segment of residues 153–172 in a protein (pdbid 1MHY) is assigned as a single α-helix by dssp while our algorithm divides it into four helices: 310-helix(154–156, yellow)–α-helix(157–163, green)–310-helix(164–166, purple)–α-helix(167–171, green). However, a careful examination of the hydrogen bond energies for these residues in fact suggests that they could also be assigned to a 310-helix even by the dssp standard (S2 Fig).

Mentions: The third column (in boldface) is the agreement on a residue basis computed as where N is the total number of residues assigned by our program, and n the number of the residues assigned by both our algorithm and either dssp or stride. The agreement on a residue basis on π-helix assignment between our algorithm and stride is very poor, only 2.9%. The 4th-10th columns present various agreements (disagreements) on a helix basis. The columns A, B, C show respectively the agreement with at most one residue difference, with the exclusion of the N-terminal residue and the exclusion of the C-terminal residue. The column A+B+C (in boldface) sums up the agreements on a helix basis. The next two columns D and E present respectively the helices assigned only by our algorithm and the helices assigned only by either dssp or stride. The last column F shows the set of helices each one has been assigned by either dssp or stride as a single helix but is divided into at least two helices by our algorithm (see Figs 4 and 5). All the data are in percentage.


An Algorithm for Protein Helix Assignment Using Helix Geometry.

Cao C, Xu S, Wang L - PLoS ONE (2015)

The illustration of the differences in assignment by our algorithm and dssp.In (a) dssp assigns the entire segment (51–65, excluding P50) in a protein (pdbid 3OY9) as an α-helix. Our algorithm divides it into two helices: 310-helix (51–52, red) and α-helix (53–63, green, purple, yellow). The α-helix stops at N63 since the Cα RMSD δ values for residue 64 and 65 are respectively 0.541, 0.431, none of them less than dmax = 0.3. In contrast, the dssp assigned helix extends to residue 65. However, as shown in the left figure, the Cα coordinates of both residue 64 and 65 deviate clearly from a helical curve. In (b) a segment of residues 153–172 in a protein (pdbid 1MHY) is assigned as a single α-helix by dssp while our algorithm divides it into four helices: 310-helix(154–156, yellow)–α-helix(157–163, green)–310-helix(164–166, purple)–α-helix(167–171, green). However, a careful examination of the hydrogen bond energies for these residues in fact suggests that they could also be assigned to a 310-helix even by the dssp standard (S2 Fig).
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4488512&req=5

pone.0129674.g005: The illustration of the differences in assignment by our algorithm and dssp.In (a) dssp assigns the entire segment (51–65, excluding P50) in a protein (pdbid 3OY9) as an α-helix. Our algorithm divides it into two helices: 310-helix (51–52, red) and α-helix (53–63, green, purple, yellow). The α-helix stops at N63 since the Cα RMSD δ values for residue 64 and 65 are respectively 0.541, 0.431, none of them less than dmax = 0.3. In contrast, the dssp assigned helix extends to residue 65. However, as shown in the left figure, the Cα coordinates of both residue 64 and 65 deviate clearly from a helical curve. In (b) a segment of residues 153–172 in a protein (pdbid 1MHY) is assigned as a single α-helix by dssp while our algorithm divides it into four helices: 310-helix(154–156, yellow)–α-helix(157–163, green)–310-helix(164–166, purple)–α-helix(167–171, green). However, a careful examination of the hydrogen bond energies for these residues in fact suggests that they could also be assigned to a 310-helix even by the dssp standard (S2 Fig).
Mentions: The third column (in boldface) is the agreement on a residue basis computed as where N is the total number of residues assigned by our program, and n the number of the residues assigned by both our algorithm and either dssp or stride. The agreement on a residue basis on π-helix assignment between our algorithm and stride is very poor, only 2.9%. The 4th-10th columns present various agreements (disagreements) on a helix basis. The columns A, B, C show respectively the agreement with at most one residue difference, with the exclusion of the N-terminal residue and the exclusion of the C-terminal residue. The column A+B+C (in boldface) sums up the agreements on a helix basis. The next two columns D and E present respectively the helices assigned only by our algorithm and the helices assigned only by either dssp or stride. The last column F shows the set of helices each one has been assigned by either dssp or stride as a single helix but is divided into at least two helices by our algorithm (see Figs 4 and 5). All the data are in percentage.

Bottom Line: The first step searches for a series of bona fide helical curves each one best fits the coordinates of four successive backbone Cα atoms.The second step uses the best fit helical curves as input to make helix assignment.The structural uniformity should be useful for protein structure classification and prediction while the accurate assignment of a helix to a particular type underlies structure-function relationship in proteins.

View Article: PubMed Central - PubMed

Affiliation: The College of Computer Science and Technology, Jilin University, Changchun, Jilin, China.

ABSTRACT
Helices are one of the most common and were among the earliest recognized secondary structure elements in proteins. The assignment of helices in a protein underlies the analysis of its structure and function. Though the mathematical expression for a helical curve is simple, no previous assignment programs have used a genuine helical curve as a model for helix assignment. In this paper we present a two-step assignment algorithm. The first step searches for a series of bona fide helical curves each one best fits the coordinates of four successive backbone Cα atoms. The second step uses the best fit helical curves as input to make helix assignment. The application to the protein structures in the PDB (protein data bank) proves that the algorithm is able to assign accurately not only regular α-helix but also 310 and π helices as well as their left-handed versions. One salient feature of the algorithm is that the assigned helices are structurally more uniform than those by the previous programs. The structural uniformity should be useful for protein structure classification and prediction while the accurate assignment of a helix to a particular type underlies structure-function relationship in proteins.

No MeSH data available.