Method for visualizing large-scale protein interaction data
View Patent ↗Disclosed is a method of visualizing large-scale protein interaction data, comprising the steps of (1) producing an initial layout by placing all nodes of protein interaction data on the surface of a sphere by increasing horizontal and vertical angles of polar coordinates; and (2) yielding a graph by iterating a process moving each node of the initial layout to an equilibrium position considering global spring forces between non-adjacent nodes as well as local spring forces between adjacent nodes on a predetermined number of times, by which large-scale protein interaction data is effectively visualized in a three-dimensional space. The method for visualization of the present invention, which is much faster than the conventional algorithms, can be used for interactive analysis, as well as provide an integrated system capable of directly visualizing query results from a protein-protein interaction database.
1. A method for visualizing large-scale protein interaction data on a display by use of a computer, comprising the steps of:
producing an initial layout by placing all proteins of protein interaction data as nodes on the surface of a sphere by increasing a horizontal angle (theta) and a vertical angle (phi) of each node in polar coordinates according to the following algorithm:
1: θ increment = 2π/√{square root over (n)}
{n: number of nodes}
2: φ increment = 2π/n
3: θ = 0, φ = 0
4: for v ε V do
5: pos[v] = (r, θ, φ)
{r: radius of a sphere}
6: θ = θ + θ increment
7: φ = φ + φ increment
8: end for; and
yielding a three-dimensional graph by iterating a predetermined number of times a process of moving each node of the initial layout to an equilibrium position and updating node positions based on spring forces between a non-adjacent nodes and the node of the initial layout as well as spring forces between an adjacent nodes and the node of the initial layout, wherein the three-dimensional graph is yielded through the following algorithm:
1: r = 1
2: repeat
3: g =0.01 · r · k 2
{k: natural spring length}
4: for v ε V do
5: D = 0
{D: displacement vector of v}
6: for u ε V, u ≠ v do
7: Δ = pos[u] − pos[v]
{pos[u]: position of node u}
8: D = D − g · (Δ/|Δ|) · (|u|/|Δ|)
{|u|: distance of u from the origin}
9: if u ε Γ(v) then
10: D = D − Δ/|Δ| · (1 − |Δ|/k)/|Γ(v)|
{Γ(v): set of nodes adjacent to v}
11: end if
12: end for
13: pos[v new ] = pos[v old ] + D
14: end for
15: r = 0.98 · r
16: until T times
{T: user-determined number or 20 by default}.
2. The method as set forth in claim 1 , wherein the process of moving each node is iterated by a number of times decided by a user.
3. The method as set forth in claim 1 , wherein the protein interaction data is yeast two hybrid (Y2H) data, Martinsried Institute for Protein Sequences (MIPS) genetic interaction data or MIPS physical interaction data.
4. The method as set forth in claim 1 , further comprising the step of simplifying the graph by limiting a distance between a node selected by a user and nodes neighboring to the node to a predetermined distance.