PRTAD: a database for protein residue torsion angle distributions

Abstract

PRTAD is a dedicated database and structural bioinformatics system for protein analysis and modelling. The database is developed to host and analyse the statistical data for protein residue level 'virtual' bond and torsion angles obtained from their distributions in databases of known protein structures such as in the PDB Data Bank. PRTAD is capable of generating, caching, and displaying the statistical distributions of the angles of various types. The collected information can be used to extract geometric restraints or define statistical potentials for protein structure determination. PRTAD is supported with a friendly designed web interface so that users can easily specify the angle types, and retrieve, visualise, or download the distributions of the angles as they desire.

Fig 1

Virtual angles. A sequence of residues i, j, k, l are located at x_i, x_j, x_k, x_l. The vectors a, b, c are virtual bonds connecting the residues, α_ijk and α_jkl are virtual bond angles, and α_ijkl is a virtual torsion angle, defined as the dihedral angle between the normal vectors n_ab and n_bc of the planes formed by a, b and b, c respectively.

Fig 2

Example angle distribution. Shown in the graph is the distribution of the virtual bond angle formed by three connected LYS residues in sequence.

Fig 3

Data structures. The above: the record of the atom in the structural database: PDB ID – ID of protein in PDB Databank; Residue – the name of the residue containing the atom; Index – the index for the atom; Atom – the name of the atom; X, Y, Z – x, y, z coordinates of the atom. The bottom: The records for the distribution of the virtual angles, one for each different type: The first one for virtual bond angles, and the second for virtual torsion angles. R1, R2, R3, R4 – residues; #αi – the number of angles in [αi, αi+1], i = 0, …, 179 (or 359).

Fig 4

PRTAD organization. This automated system could generate and process the data dynamically. The system is implemented in MySQL and perl. The user could access freely the database at http://www.math.iastate.edu/prtad. It requires specifying and inputting the angle type and then the user could choose to view the graph of distribution function as well as download the related results.

Fig 5

PRTAD input selections. A user needs to first decide the database to be searched, bond or torsion angle database, and then specify the types of the residues (20 possibilities) in the sequence.

Fig 6

Graphics display. The distribution of the angles of the specified type is displayed in a graph. The angle range is up to 360°, and the size of each angle interval (bin) is 1°. (a) Torsion angle for AQHF. (b) Torsion angle for RQGL. (c) Bond angle for KKK.

Fig 7

NMR structure of prion fragment 1OEH. The fragment extends from residue 61 to 68 in prion protein, with a sequence HGGGWGQP. The structure was determined by NMR with an ensemble of 20 structures available in PDB.

Fig 8

Angle for GGG in 1OEH. (a) The distribution of the angle calculated by PRTAD. (b) The angles in the 20 NMR structures.

Fig 9

Structure of p53. Methylation may occur at either K370 or K372. The sequence between the two sites is KSKK (picture obtained from [8]).

Fig 10

Angle for KSKK in p53. The distribution of the angle calculated has two separate peaks corresponding to two possible conformations of the fragment.