Nucleic acid – protein interaction
In NPIDB, a number of forces connecting nucleic acid and protein molecules to each other are caluclated.
Hydrogen bonds are computed by a homemade program called "hb_angles". The principles of its algorithm are as follows:
- hydrogen bonds are formed by closely positioned atoms of oxygen and nitrogen;
- values of atom-to-atom distances and other geometrical features (angles formed by interacting atoms and their neighbors in molecules) that are typical for hydrogen bonds are extracted from the existing structures;
- typically, the distance between centers of bonded atoms should be less than 3.5 Angstroms, but sometimes (e.g., for structures with relatively low resolution) the distance threshold may be set to higher values, up to 3.9 angstroms. At the moment only hydrogen bonds with the distance less than 3.7 are included into the database;
all oxygen and nitrogen atoms are divided in two types: atoms that have one covalent bond with heavy (not hydrogen) atoms are called atoms of type I, atoms that have two covalent bonds with heavy atoms are called atoms of type II;
- the main geometrical values of a hydrogen bond are: the distance between the centers of the bonded atoms and two theta angles. For a type I atom, the theta angle is the angle between two directions: the first is the direction from its covalently bonded neighbor to it, the second is the direction from it to the other atom of the hydrogen bond. For a type II atom, the first direction is the mean between two directions from two covalently bonded atoms to the atom;
- for a pair of two oxygen atoms, the most typical distance of a hydrogen bond is 2.7 angstroms (between the centers of the oxygen atoms). For a pair of two atoms one of which is oxygen and one is nitrogen, the most typical distance is 2.9 angstroms. Hydrogen bonds between a nitrogen atom from nucleic acid and a nitrogen atom from protein are very rare; the most typical distance is hard to extract from data, but it is conventionally set to 3 angstroms.
for a type II atom, the most typical theta angle is 0. For a type I atom, the most typical theta angle depends on the chemical nature of the atom (1.15 radian for NZ atom of lysine or OG atom of serine, 1.05 radian for OH atom of tyrosine). At the moment, the most typical theta angles for all type I atoms are set to 1.15 (this will be improved soon, we need some additional research);
- the program calculates a conditional value called "power" that indicates how typical is a given configuration of atoms for hydrogen bonds. The value of power is between 0 (absolutely non-typical configuration) to 1 (the most typical configuration). At the moment, only hydrogen bonds with the power greater than 0.1 are included into the database.
The formula for the power:
where d is the distance between centers of donor and acceptor atoms; θnuc and θprot are theta angles,
where dopt is the most typical distance;
for atoms of type I (0.4 is the cosine of 1.15 radian, the most typical theta angle);
for atoms of type II.
Water bridges (or water-mediated contacts) are computed by the same program "hb_angles", which computes hydrogen bonds. A water bridge between an atom of protein and an atom of nucleic acid is detected if there exist a water molecule that forms hydrogen bonds with both these atoms.
The power of a water bridge is calculated by the formula:
Here dnuc and θnuc are the distance and the theta angle for the bond connecting the water molecule with the nucleic atom, dprot and θprot are the distance and the theta angle for the bond connecting the water molecule with the nucleic atom. The formulas for pd and pa are the same as in the case of hydrogen bonds. There are only two theta angles, because the oxygen atom of water has no covalent bonds with other heavy atoms, thus for a bond between water and protein or nucleic atom, the theta angle from the side of water is not defined.
At the moment, only water bridges with the power greater than 0.1 are included into the database.
The hydrophobic interactions are computed by the program CluD (a description of its algorithm see at http://mouse.belozersky.msu.ru/npidb/html/help.htm ). The output of the program is a table of atoms that are detected in hydrophobic clusters, with numbers (id's) of the clusters.