RING - Residue Interaction Network Generator
|Quick Help||Output||References||Examples||Method||Cytoscape||File format|
This type of network defineds a link between a pair of
non-adjacent residues if the distance between their alpha
carbons is ≤ the threshold value. This
threshold can be changed by introducing a new value in the form
cutoff. The definition implies that only one
connection between a pair of residues can be traced (dashed red
line in the image below).
This type of network can generate 11 different types of edge attributes and 12 node attributes. Once loaded into CYTOSCAPE, these attributes will be visible in the data panel when you select one or more connections or nodes. The edge attributes are:
The node attributes are
For explanations about the meaning of the attributes listed see the explanation paragraphs for the respective methods.
A network of this type can be useful for coarse inspection; attributes only help you to remember what types of strong interactions exist between the residues in contact, without wanting to specify anything else.
In this type of network a link is defined if the distance (in
Å) between a pair of non-adjacent residues and their
closest atoms is ≤ than the threshold value. The
threshold can be changed by introducing a new value in the form
cutoff, (dashed red line in the image).
If the closest atoms determining a link between a pair of residues form a hydrogen bond, then the connection is classified as a hydrogen bond (dashed red line in the image across). However, unlike the previous case, if the same pair of residues are found involved in other hydrogen bonds that do not involve the closest atom, more directed links are added (dashed blue line in the image across). This means that the couples involved in several hydrogen bonds will be represented by pairs of nodes with multiple links.
For this type of network there are 29 different types of edge attributes stored in as many files and 12 node attributes. Once loaded into CYTOSCAPE, these attributes will be visible in the data panel when selecting one or more connections. The edge attributes are:
The node attributes are:
For explanations about the meaning of the attributes listed see the explanation paragraphs for the respective methods. This type of network useful when one wants to analyze interactions that are present within a protein on the base of the distances between residues.
To generate this type of network RING uses PROBE to identify the van der Waals interactions between the atoms of each pair of considered residues and defines the two residues as in contact if between any of their atoms exists at least one van der Waals interaction.
In this case, as many links as the van der Waals interactions that are identified between the same pair of residues defined in contact are added. RING subsequently identifies all the other types of noncovalent interactions in the constructed network, and adds the necessary connections for them.
For this type of network there are 30 different types of edge attributes stored in as many files and 12 node attributes. Once loaded into CYTOSCAPE, these attributes will be visible in the data panel when selecting one or more connections. The edge attributes are:
The node attributes are:
For explanations about the meaning of the attributes listed see the explanation paragraphs for the respective methods. This type of network is very useful when one wants to analyze the actual number of interactions present within a protein.
RING assigns to each established connection a
weight, with the exception od peptide bonds when these are generated. This weight can be used to calculate
some topological parameters for the generated network (like
shortest path betweenness centrality, closeness centrality,
shortest path degree centrality...) using some CYTOSCAPE
plug-ins available on the web.
Weights for the closest atom networks
The attribute weight generated by RING for the closest atoms network correspond to the approximate average free energy of each interaction and represent the average values used in the literature arising from both simulations and experimental data. These values are:
For the disulfide bridge free energy the value used is the dissociation enthalpy of this covalent bond. For simple interactions (i.e. between closest atoms among which there is no other classified interactions) we consider the average attractiveness component of the van der Waals force that exist between any two residues.
Weights for the van der Waals networks
The attribute weight generated by RING for
the van der Waals networks is the same generated for the closatom network (see above). In addition the edge attribute van der Waals contact score is used as weight of the van der Waals interactions. The van der Waals contact score is generated for every van der Waals interaction as follows:
|Weights for the C-alpha networks|
For the C-alpha network the attribute weight generated by RING is = 1 for all interactions. This is because in such type of network, atoms that determine the interactions (C-alpha) can be affected at most by the van der Waals force which is very small because generally, there is a large distance between pairs of them.
HBexplore provides 2 geometric criteria to identify potential hydrogen bonds.
The first and simplest is based on constraints with respect to distances between the acceptor atom A and the hydrogen atom H, called d1, and between donor atom D and acceptor atom A called d2. The following are calculated: the angles between the bond -D-H and d1, called α; the link between A-A1- (where A1 is the atom which is covalently linked to the acceptor) and d1, called γ; the angle between A-A1 and d2, called β. If the acceptor atom A is part of a ring, then it will have two neighbors, A1 and A2. In this case the atom which is covalently linked to A (i.e., Am) is lying in the middle point along the line joining A1 and A2, (see the image across). For donor atoms with sp3 hybridization the distance H-A is first determined but only after the other geometrical parameters are calculated. In RING the threshold used is the same as in HBexplore: d1 < 2.5 Å, d2 < 3.9 Å, α > 90°, β > 90° γ > 90°.
The second criterion is based on the directionality of the hybrid
orbitals and it arises from the observation that many hydrogen
bonds show a clear directionality. This comes from the
overlap of hybrid orbitals because while the s orbital of
the donor has a spherical symmetry, the hybrid orbital of the
acceptor atom in sp2 or sp3
configuration has an angle of 120° and 109° respectively. In
HBexplore, the orbitals are treated as vectors whose
direction is inferred from the atoms surrounding the acceptor. To
constrain the angle DHA, the angle between H-A and
the direction of acceptor hybrid orbitals (Hyb) for which
there must be at least one with an angle less than 60° are also
taken into account (see the image across). This criterion considers
as donor atoms only O, N and sulfur S while
the C is not covered. RING uses the
standard threshold used by the HBexplore:
d1 < 2.5 Å,
d2 < 3.9 Å, α > 90°,
β > 90°, ε < 60°.
RING investigates on the presence or absence of a
salt bridge when a residue is usually negatively charged at
physiological pH. Asp or Glu, is usually set in salt contact
with a usually positively charged residues like Arg, Lys or His.
RING also calculates the angle ε of the
bridge, in degrees, defined as the angle formed between the
two lines connecting the C-alpha with the mass center of the
charged group, respectively in each pair of residues (see ρ in image
a pair of residues defined as being in contact, RING
checks if they both belong to the category of aromatic amino acids
(His, Tyr, Trp, Phe). Then RING states if a
π-π interaction exists between them if
there is at least one pair of atoms belonging to one of the
two side chains at distance < 6 Å apart (see
image across). This default threshold value can be changed by
entering a new value in the field π-π
minimum distance. RING subsequently tries to
determine the reciprocal position of the two aromatic rings
considered using a combination of spatial and rotational
RING also defines three rotational orientations:
These rotational orientations are defined on the basis of the dihedral angle θ (the angle in degrees between the planes of the two rings). The rotational and spatial orientations are combined to define four spatial configurations of the two rings:
The constraint on the minimum distance between side chains that identifies a π-π interaction was derived empirically from the analysis of a large group of protein structures. It has to be noticed that there are more accurate methods to determine this interaction but these are computationally more expensive and only small improvements are reported.
As for the remaining constraints on n, p and θ the solution here is widely adopted in the literature to evaluate the reciprocal position of aromatic systems.
For this type of interaction the links are not directed.
RING searches for π-cation interactions when two residues in contact are respectively an amino acid positively charged at physiological pH (Arg or Lys) and an amino acid with aromatic side chain (Phe, Tyr, Trp). His, although possessing an aromatic side chain, is not considered because it can participate in this type of interaction both as cation and as a π-system, depending on its protonation state.
To identify a π-cation interaction two conditions must be met:
The default thresholds for the distance and angle are empirically derived. These constraints intend to select those cations whose positive charge lies above or below the ring area, where the ring surface potential is negative. This criterion is approximate but sufficient for our purposes and not very expensive in terms of computational time.
Moreover, given the different intensity of interaction between Arg and a π-system according to the orientation of the guanidinium ion, RING divides this interaction into three categories. The catagories depend on the dihedral angle θ in degrees defined between the plane of the aromatic ring and the cationic guanidine group:
The rules to calculate the mass center of charged group
in Lys and Arg are the same as in the method used to search
for salt bridges (see
above), i.e. for Nζ Lys and
of the guanidine group in Arg. For the calculation of mass centers for
the aromatic systems the same rules are applied as defined for
π-π interactions (except in the
case of His because it is not considered (see above)).
Being a covalent bond between two atoms, disulfide bridges
present a constant distance. RING uses this
type of constraint to identify disulfide bridges between
two Cys residues being in contact.
correct orientation of Asn and Gln side chain -NH2
against O atoms is very important when these residues are
involved in hydrogen bonds. This is especially true when these hydrogen bonds occur in the active site of a
protein or when examining the network of hydrogen bonds in a
The program PROBE allows one to evaluate atomic packing within molecules. PROBE reads atomic coordinates from protein databank (PDB) format files and generate contact dot-lists where atoms are in close contact. Alternatively, the packing information can be quantified and displayed in a table listing scores for van der Waals interactions, H-bonds and atomic overlaps ("clashes"). Essentially, the approach is to place a very small probe (typically of radius 0.25Å) at points along the van der Waals surface of a selected set of atoms and determine if this probe also contacts atoms within a second "target" set. So PROBE provides a flexible method for selecting the source and target atoms.
Analysis of molecular contact surfaces by PROBE requires that ALL atoms are considered for this, before using PROBE, the PDB file must be submitted to the program REDUCE to add hydrogens to the coordinate file (see above).
The non-overlapping van der Waals contacts are quantified by an error-function weighting, which awards close contacts a higher score than distant or significantly overlapping ones, but slight overlaps are still favorable in net effect. The combined score is a weighted sum of the weighted non-overlapping van der Waals contacts, the volume of hydrogen bonds and the volume of overlaps. The PROBE program summarizes these scoring data for all parts of an entire structure and can output a file with information for every atom or residue.
RING runs PROBE on the PDB file containing coordinates of all hydrogen atoms added by REDUCE and uses the contact list generated by it to get information about the van der Waals interactions between pairs of residues defined in contact. RING also catch the scores of each interaction to formulate a weighted score for each of them, depending on the site where interaction occurs (see above).
The degree of a node is its total number of
edges (i.e. the number of neighbors). When generated, peptidic bonds are not used to compute node degrees. Taking into account the
important role that could be played in a protein by amino acids
with high connectivity in the network (the so-called hubs),
RING produces a file containing a multiple
sequence alignment in FASTA format, built on the amino
acid sequence of the protein being examined, in which each
sequence is filtered according to the increasing
degree of the corresponding nodes.
The multiple sequence alignment can be easily loaded
and manipulated in JALVIEW,
which is a useful tool to visually capture which amino
acids act as hubs according to the network (see output
page). These residues are those which persist longer in the
multiple sequence alignment (i.e. they are more evolutionary
conserved). Along with the alignment RING also produces
a text file with extension
DSSP (The Dictionary of Protein Secondary Structure) is a commonly used program to describe proteins secondary structure. Among many other features it also provides solvent accesibilty (Å2). DSSP assigns secondary structure based on hydrogen bonding patterns. Hydrogen bonds are defined by DSSP using an electrostatic model. DSSP assigns a charge of ± q1 = 0.42e to the carbonyl carbon hydrogen (+) and oxygen respectively (-), and charges of ± q2 =0.20e to the amide nitrogen (-) and hydrogen (+), respectively. The electrostatic energy (E) is:
E = q1 q2 f ( 1/rON + 1/rHC' - 1/rOH - 1/rNC'} ).
where f = 332 kcal/mol and rXX are the respecitve distances. Hydrogen bonds only exists if E is less than -0.5 kcal/mol for a given pair donnor acceptor. The eigth secondary structure classes are:
DSSP computes solvent accesibility by approximating the protein surface which could be in contact with an spherical water molecule using an algorithmic called by DSSP authors "geodesic sphere integration" (see the DSSP paper for a more detailed description). Solvent accesibility values are normaliced to provide relative solvent accesibility. Normalicing values are taken from Miller et al.
RING provides the posibility of directly generating sub-networks containing only residues above or below a user defined relative solvent accesibility threshold. In this case, all the output files only contain interactions and attibues of those amino acids acomplishing the selected criteria. The by defoult threshold of 25% is the more commonly used to distinguish between buried and solvent exposed residues. This posiblity can be conbined with sub-networks containg only residues with a conservation above a certain threshold, i.g. to generate a sub-network containing only exposed amino acids with 50% conservation (see below).
RING produces a multiple sequence alignment in FASTA format running PSI-BLAST against the UniRef90 sequence database. In the Complex form, the user can specify the number of PSI-BLAST iterations and the minimum E-value to include sequences in the alignment. All the other PSI-BLAST parameters are the by default values. In RING, residue conservation is defined as the percentaje of sequences in the multiple sequence alignment in which an amino acid has the same identity as in the protein being studied. Highly conserved residues play an important role in the function and structure of proteins.
RING provides the posibility of directly generating sub-networks containing only residues above a user defined conservation threshold. In this case, all the output files only contain interactions and attibues of those amino acids acomplishing the selected criteria. This posiblity can be conbined with sub-networks containg only residues with a relative solvent accesibility above a certain threshold, i.g. to generate a sub-network containing only exposed amino acids with 50% conservation (see above).
RING produces an energy evaluation performed by FRST. FRST computes a per-residue energy profile of a protein structure. This profile is a composite score made of RAPDF pairwise potential, solvation potential, bakbone hydrogen bonds and torsion angle potential. For more a more detailed description see FRST reference.
RING also evaluates the protein structure with TAP score. TAP evaluates the local torsion angles of a protein structure, producing a single value per residue which provides information on the likelihood that the amino acid will satisfy particular experimental quality criteria. For more a more detailed description see TAP reference.
The temperature factor or B-factor can be thought of as a measure of how much an atom oscillates or vibrates around the position specified in the model. Atoms at side-chain are expected to exhibit more freedom of movement than main-chain atoms, and this movement amounts to spreading each atom over a small region of space. Diffraction is affected by this variation in atomic position, so it is realistic to assign a temperature factor to each atom and to include the B-factor among attributes available.
For this RING reports for each residue in the network, the value of the C&alpha B-factor as found in the PDB file.
From the temperature factor of the C&alpha carbons it's possible to learn what portion of the main chain has most freedom of movement and gain some insight into the dynamics of our largely static model.
The occupancy of atom j is a measure of the fraction of molecules in the crystal in which atom j actually occupies the position specified in the model. It is a normalized value between 0 and 1 in which the value 1 indicates that the occupancy of all atoms in the same position is precisely identical. For example, if the two conformations occur with equal frequency, then atoms involved receive occupancies of 0.5 in each of their two possible positions.
Given that occasionally two or more distinct conformations are observed for small regions of the protein, the occupancy is included among the parameters refined. RING reports the value of occupancy of the C&alpha, as found in the PDB file, for each residue shown in the network.
In this way the user get an estimation of the frequency of alternative conformations of a given residue in the main chain of the protein, obtaining some additional information on the dynamics of the molecule.
Mutual Information can be used to estimate the strength of the coevolutionary relationship between two residue positions in a multiple sequence alignment (protein family) . It value is given by the following formula: