A) PDB code 1GPI. |
B) PDB code 1EZG. |
C) PDB code 1JUB. |
D) PDB code 1H16. |
|
What is Beta-Spider? |
Beta-Spider is an algorithm that computes the beta-sheet domains of a given protein using its 3D atomic coordinates. It has the same purpose as DSSP [4], DSSPcont [5], STRIDE [6], DEFINE [7], P-Curve [8], P-SEA [9], xtlsstr [10], STICK [11] and VoTAP [12]. The algorithm is implemented in a program that is available on this web page. The program takes for input a PDB file and replaces all the SHEET records in that PDB file with those calculated by the algorithm. Thus, the output of the program is readily available to analysis to any automated protein data processing from the PDB file.
The assignation of secondary structure elements from atomic coordinates follows mainly two principal routes. The first approach is to use purely geometrical factors (Ca-Ca distances, phi/psi angles, etc.) like in DEFINE, P-Curve, P-SEA, xtlsstr, STICK and VoTAP. The other approach is to use template matching on hydrogen bonds, like in DSSP and DSSPcont. Elsewise, STRIDE is an hybrid approach that uses both the geometry of the chain via the usage of the phi/psi angles and also some pattern matching on H-bonds.
Beta-Spider uses a different approach by considering all the stabilizing forces at play in the beta-sheet phenomenon. Thus, not only the C=O...H-N hydrogen bonds are considered but also the C=O...C=O electrostatic dipoles [14,15] and bifurcated H-bonds C=O...Ha-Ca [16-21]. Although Beta-Spider uses some geometrical factors, to make sure that the side-chains of the beta-sheet partners are pointing in the same direction, it does not attempt to match any atomic coordinates on a given reference template, nor use the phi/psi angles, nor simplify the main chain into a mathematical abstraction. Instead of using the three [4] or five [6] H-bonding templates, the program proceeds to a full inter-chain non-bonded energy calculation, thus taking into account the standard C=O...H-N hydrogen bonds. In particular, the program shows that there are many more H-bond patterns that those already used in [4] or [6] and that it is not a better definition of an H-bond that saves the day (for example, in [6] the authors use a much more elaborate hydrogen bond calculation).
Our motivation to extend DSSP's annotation of beta-sheets comes from the fact that automated protein fold classification would not have worked given the beta-sheets identified by DSSP in Figure 1, among others. We have not based our method on any assuptions regarding the phi/psi angles nor the hydrogen bonding patterns found in beta-sheets.
|
ONLINE ANALYSIS |
|
PDB SHEET RECORDS |
The SHEET record in the PDB file format is not appropriate to describe unambiguously protein beta-sheets. The description of bifurcated beta-sheets leads one to think that there are many sheets while, in fact, there is only one sheet under description. Also, the registration fields does not determine uniquivocally all inter-strand beta-sheet partners. For example, consider the case of a bulged residue.
The following beta-sheet description for an 8-stranded beta-barrel (PDB code 1TIM):SHEET 1 A 9 PHE A 6 TRP A 12 0 1TIM 119 SHEET 2 A 9 ASP A 36 GLY A 42 1 O ASP A 36 N PHE A 6 1TIM 120 SHEET 3 A 9 GLY A 60 ALA A 63 1 N GLY A 60 O VAL A 39 1TIM 121 SHEET 4 A 9 ALA A 89 LEU A 93 1 N ILE A 92 O ALA A 63 1TIME 6 SHEET 5 A 9 GLY A 122 GLU A 129 1 O GLY A 122 N VAL A 91 1TIM 123 SHEET 6 A 9 LYS A 159 VAL A 167 1 O LYS A 159 N VAL A 123 1TIM 124 SHEET 7 A 9 ARG A 205 GLY A 209 1 O ARG A 205 N LEU A 162 1TIM 125 SHEET 8 A 9 ASP A 227 VAL A 231 1 O GLY A 228 N TYR A 208 1TIME 7 SHEET 9 A 9 PHE A 6 TRP A 12 1 N PHE A 7 O ASP A 227 1TIME 8is now outputted by Beta-Spider as:
SHEET 1 001 8 PHE A 6 TRP A 12 0 SHEET 2 001 8 THR A 37 ALA A 43 +1 N VAL A 40 O GLY A 10 SHEET 1 001 8 PHE A 6 TRP A 12 0 SHEET 8 001 8 VAL A 226 GLY A 232 +1 N PHE A 229 O GLY A 9 SHEET 2 001 8 THR A 37 ALA A 43 0 SHEET 3 001 8 ILE A 59 ASN A 65 +1 O GLY A 60 N VAL A 39 SHEET 3 001 8 ILE A 59 ASN A 65 0 SHEET 4 001 8 ALA A 88 HIS A 95 +1 N ILE A 92 O ASN A 65 SHEET 4 001 8 ALA A 88 HIS A 95 0 SHEET 5 001 8 GLY A 122 GLU A 129 +1 N ILE A 124 O LEU A 93 SHEET 5 001 8 GLY A 122 GLU A 129 0 SHEET 6 001 8 LYS A 159 PRO A 166 +1 O GLU A 165 N ILE A 127 SHEET 6 001 8 LYS A 159 PRO A 166 0 SHEET 7 001 8 ARG A 205 GLY A 209 +1 N ARG A 205 O LEU A 162 SHEET 7 001 8 ARG A 205 GLY A 209 0 SHEET 8 001 8 VAL A 226 GLY A 232 +1 O ASP A 227 N ILE A 206
first of all, notice the length of the description of the beta-sheet by the program Beta-Spider; 16 lines instead of only 9 lines for the original description in the PDB file. The reason of that is that Beta-Spider describes sheets from the pairs of strands in contact. Thus, two SHEET records are produced for each pair of strands in contact; the first strand serves as the reference (i.e. has orientation 0) while the second strand is described relatively to the first one (with the +1/-1 strand orientation). This is an elegant alternative to the PDB format proposition for bifurcated beta-sheets. Here, Beta-Spider outputs the proper number of strands in the sheet, 8, instead of 9, as indicated in the original PDB file. Each strand as a unique strand identification number (columns 8-10) and is properly assigned in the sheet description generated by Beta-Spider, as opposed to the original sheet description in which the strand of residues 6 to 12 as two different strand number, namely 1 and 9. The residue pair in the registration fields is chosen as the pair that has the lowest (i.e. better) H-bond energy.
|
Catalogs |
Here are two catalogs of beta-sheets compiled by DSSP and Beta-Spider. They are built from a culled PDB Select 25 protein chains list, which has been established by the Pisces server (at http://dunbrack.fccc.edu/PISCES.php) [22]. This list is also provided for download.
|
REFERENCES |
|
CONTACT |