Base pair classification (under construction)
© 2000, LBIT

Pairing tables
A-A pairings 10
A-C pairings 15
A-G pairings 21
A-U pairings 15
C-C pairings
8
C-G pairings 19
C-U pairings 12
G-G pairings 14
G-U pairings 16
U-U pairings
5

Currently, there are at least two accepted classification of base pairing patterns. The first one uses a simple denomination based on the H-bonds that are formed in each pairing. W. Saenger [1] originally used this denomination for pairings involving two or more H-bonds. D. Gautheret et al. [2] later added pairings involving only one H-bonds. The RNA modeling software Mc-Sym currently uses this classification.

The second classification, or geometry based nomenclature, comes from N.B. Leontis & E. Westhof [3] and employs a naming scheme based on the chemical "faces" involved in the pairings as well as on the relative orientation of the sugar to ribose bonds.

Unfortunately, none of these representations captures all the information needed to establish a one-to-one correspondence between a particular pairing and the associated name. The advantages and disavdvantages of each representation are indicated below.

As a part of the elaboration of the Madison format for secondary structure, we wish to design a new classification that will uniquely represent each type of pairing interaction and let a user be as specific as wanted when naming a pairing.

The set of pages on the left contains all identified base pairings, as well as the corresponding names and number of occurrences in the PDB structures.

Isosteric pairs are computed using the distance metric described in [4] using a distance cutoff of 1.5 Å.



Characteristics of the MC-SYM nomenclature:
  • + There is a one-to-one correspondance between the name of the pairing and the set of H-Bonds involved.
  • - Cannot discriminate between base flipping of one H-Bond pairings.
  • - The names are hard to assign without an automated method.
  • - In pairings involving two identical bases, the name does not discriminate the direction of interaction (ex: A-A type V).
Characteristics of the geometry based nomenclature:
  • + The cis/trans values discriminate between base flipping of one H-Bond pairings.
  • + The names are easy to assign.
  • - Two different pairings (involving different H-Bonds but the same "faces") can have the same name. (Maybe it is not really a disadvantage for MC-Sym).
  • - There is no indication of the involved H-Bonds.
  • - In pairings involving two identical bases, the name does not discriminate the direction of interaction (This is solved by considering the order of the faces in the name, a A-A Wc/Hoogst is different from a A-A Hoogst/Wc).
Proposition:
  • Use the Leontis nomenclature in MC-Sym and in the Madison Format with an ordering of the properties based on the lexicographical order of the base
  • DONE: Add a fourth parameter to the Leontis Format that discriminate between pairings that involve the same chemical "faces" (still to define)
  • DONE: Find a way to talk about pairings that are different when a reference base is fixed (for example A-A type V)
  • DONE: Implement this classification in MC-Sym and Annotate
  • Implement a method to specify the chemical groups involved in a pairing in MC-Sym.

[1] W. Sanger, Principles of Nucleic Acid Structure, Springer-Verlag, 1984.
[2] D. Gautheret and R.R. Gutell (1997) Inferring the conformation of RNA base pairs and triples from patterns of sequence variation, Nucl. Acids Res., 25(8):1559-1564.
[3] N.B. Leontis and E. Westhof (1998) Conserved geometrical base-pairing patterns in RNA, Quaterly Review of Biophysics, 31(4):399-455.
[4] P. Gendron, S. Lemieux and F. Major (2000) Quantitative Analysis of Nucleic Acid Three-Dimensional Structures, submitted to J. Mol. Biol..