Transcript Poster - Computer Science and Engineering

Discovering the Correlation Between
Evolutionary Genomics and Protein-Protein Interaction
Rezaul Kabir ([email protected]) and Brett Thompson ([email protected])
Department of Computer Science, University of North Texas
Introduction
Finding the protein interactions that are responsible for cellular
operations has become one of the main goals of proteomics and
computational biology. The prediction of protein-protein interactions
is a computational intensive problem in bioinformatics. Studies have
shown for an evolutionary genomics comparison that the number of
protein-protein interactions a protein has negatively correlates with
their rates of evolution. Currently there are new and powerful
experimental techniques used to discovery these interaction
networks. With the current methods being far from perfect, we will
show some of the current and effective techniques being used today
and how those techniques will help researchers and scientists
uncover the evolution of genomes.
First Methodology: Using
Phylogenetic Trees to Predict
Protein Interaction with the
Partial Correlation Coefficient
A new method to predict protein-protein interaction from
evolutionary information using partial correlation coefficient
extracts direct protein interactions unlike Peason’s correlation
coefficient, which only gives indirect interactions between
proteins. The partial correlation coefficient uses the comparison
of phylogenetic trees of proteins to predict physical protein
interactions [3].
Second Methodology: The Use of
Phylogenetic Trees as Indicators
One method to help open up the possibilities of searching for interaction partners
between proteins in a large collection of complete genomes and proteins is to use
the comparison of the evolutionary distances between the sequences of the
associated protein families. This comparison is based on the observations of
correspondence between phylogenetic trees of associated proteins. The method
measures the similarity between trees as the correlation between the distances
matrices used to build the trees based on the mirror tree method. The mirror tree
method assumes that functionally correlated proteins evolve in a correlated form
(Fig. 2). With Peason’s correlation coefficient based on phylogenetic trees, the
mirror tree method is able to evaluate the intensity levels between correlated
proteins [1].
Fig. 2: This figure shows the scheme of the Mirror
Tree method. This method reduces the initial multiple
sequence alignments of the two proteins, which
leaves only sequences of the same species. The trees
constructed from these reduced alignments will have
the same number of leaves and the same species in
the leaves. From the reduced alignments, the matrices
are constructed which contain the average homology
for every possible pair of proteins. Such matrices
contain the structure of the phylogenetic tree. Finally,
a linear correlation coefficient evaluates the
similarity between the data sets of the two matrices
and implicitly the similarity between the two trees
[1].
Third Methodology: Identifying and
Classifying Protein Interaction
Interfaces with InterPare
InterPare (http://InterPare.net) is a large-scale protein domain
interaction interface database. The interface consists of both interchain (between chains) and intra-chain (within chains). The three
methods InterPare uses to detect protein-protein interaction are the
geometric distance method (PSIMAP), Accessible Surface Area
(ASA), and the Voronoi diagram. There are visual tools to display
protein interior, surface, and interaction interfaces and statistics of
the amino acid propensities of queried protein according to its
interior, surface, and interface region (Fig. 3 Left). InterPare makes
searching and looking up of protein-protein interaction easy and
convenient [2].
Fig. 3 (left): The figure shows the protein
structure with respect to their geometrical
region. This is an example of a 3D structure
(SCOP id: d1a25a_) which corresponds to a
schematic diagram. It shows the three areas of
a domain (red: protein surface, blue: protein
interior, filled-in space model: interaction
interface). Interface regions are represented as
a space-fill model to distinguish them from
other regions [2].
Fig. 4 (bottom): This diagram represents the
interior, interface, and surface of longitudinal
section of a protein domain [2].
80
70
Final Thoughts and Conclusion
60
50
80
40
30
20
20
10
0
1
2
Fig. 1: This graph compares the accuracy of the Peason’s coefficient and the partial correlation
coefficient using the first five top-ranking predictions. In column 1, the Peason’s correlation
coefficient has an accuracy of 20% (1/5). In column 2, the partial correlation coefficient has an
accuracy of 80% (4/5) [3].
Studies have shown for an evolutionary genomics comparison that the number of protein-protein interactions a protein has negatively correlates with their rates
of evolution [4]. In order to find the correlation between evolutionary genomics and protein-protein interactions, we must compile several data sets to show any
significant relationship, since the correlation can not be shown with a small set of protein-protein interaction [5]. With a large interaction data set, researchers
able to assess the quality of the data set through the correlation between protein interaction and evolutionary rate using a simple genomic sequence
comparisons statistically.
References
[1] Pazos, F. and Valencia, A., Similarity of phylogenetic trees as indicator of protein-protein interaction, Protein Eng., 14(9):609–614, 2001.
[2] S. Gong, C. Park, H. Choi, J. Ko, I. Jang, J. Lee, D. Bolser, D. Oh, D. Kim, and J. Bhak, Using Interpare, a protein domain interaction interface database, to identify and classify protein interaction interfaces.
[3] Toh , H. and Kanehisa, M., Predicting protein-protein interaction from phylogenetic trees using the partial correlation coefficient.
[4] H. Fraser, A. Hirsh, L. Steinmetz, C. Scharfe, and M. Feldman Evolutionary Rate in the Protein Interaction Network, Science Magazine., 296:750-752, 2002.
[5] H. Fraser, D. Wall, and A. Hirsh A Simple Dependence Between Protein Evolution Rate and the Number of Protein-Protein Interactions, BMC Evolutionary Biology., 3:11, 2003