Biological processes in the live cycle of a cell such as proliferation, signal transduction or apoptosis are shaped by proteins interacting specifically with each other and building more or less transient complexes. To understand these processes, determining the underlying protein interactions is of vital importance.
In recent years, high-throughput methods such as yeast two-hybrid (Y2H) and tandem affinity purification (TAP)
have made it possible to determine protein interactions on a large scale.
As a consequence, species-specific interaction networks have been and are still determined for a range of organisms.
We develop methods and provide assistance for the analysis of large-scale interaction networks determined, e.g. by yeast two-hybrid,
the identification of protein complexes and their substructures from affinity purification experiments and the extraction of
known interactions from literature.
Analysis of experimental interaction networks
Analysis of intraviral protein-protein interactions of the SARS coronavirus ORFeome.
For severe acute respiratory syndrome coronavirus (SARS-CoV),
900 pairwise interactions were tested by yeast-two-hybrid matrix analysis,
and more than 65 positive non-redundant interactions, including six self interactions, were identified.
Analysis of network statistics showed that despite high clustering coefficients the SARS
interaction network is not higher clustered than expected at random.
Based on currently known and predicted host-virus interactions,
a joint virus-human network was derived in which the viral part of the network
appears to be separated from the main host network.
However, this may be due to the small number of virus-host interactions
identified so far for SARS.
To better understand the role of the intraviral protein interactions it
is necessary to gain more knowledge on the SARS-CoV with it's host during infection.
Comparative Interactomics reveal evolutionary conserved Herpesviral Protein-Protein Interaction Networks
Herpesviruses constitute a family of large DNA viruses widely spread in vertebrates and
causing a variety of different diseases.
In this study, we systematically screened the interactomes of three
herpesvirus species, Herpes simplex virus type 1 (HSV-1), murine Cytomegalovirus (mCMV) and Epstein Barr Virus (EBV)
to previously published interactomes for Varicella Zoster Virus (VZV) and Kaposi?s sarcoma herpesvirus (KSHV).
We were able to identify a core set of highly conserved protein interactions.
Interactions were conserved between orthologous proteins despite generally low sequence similarity,
suggesting that function may be more conserved than sequence similarity.
By combining interactomes of different species we were able to systematically address
the low coverage of the Y2H system and to extract biologically relevant interactions
which were not evident from single species.
Analysis of network topology
Inferring topology from clustering coefficients in protein-protein interaction network
We investigated the effect of limited sampling on average clustering coefficients
and how this can help to more confidently exclude possible topology models for the
complete interactome. Both analytical and simulation results for
different network topologies indicate that partial sampling alone lowers
the clustering coefficient of all networks tremendously.
Furthermore, we extend the sampling model by also
including spurious interactions via a preferential attachment process.
Simulations of this extended model show that the effect of wrong interactions
on clustering coefficients depends strongly on the skewness of the
original topology and on the degree of randomness of clustering coefficients in the corresponding networks.
Influence of degree correlations on network structure and stability in protein-protein interaction networks
We analyzed a range of experimentally derived interaction networks to understand
the role and prevalence of degree correlations in PPI networks.
Furthermore, we investigated how degree correlations influence the structure of
networks and their tolerance against perturbations such as the targeted deletion of hubs.
We found that negatively correlated networks are fragmented into
significantly less components than observed for positively
correlated networks. On the other hand, the selective
deletion of hubs showed an increased structural tolerance to
these deletions for the positively correlated networks.
This results in a lower rate of interaction loss in these networks compared
the negatively correlated networks and a decreased disintegration rate.
Interestingly, real PPI networks are most similar to the randomly correlated
references with respect to all properties analyzed.
Recent advances in high-throughput technologies allow the large-scale determination of proteins interaction directly or indirectly with another protein in vivo. From these interaction complexes of proteins can be determined which bind together to perform a biological function. So far, approaches to this problem rely on previously reported information on protein complexes to predict new complexes. We investigate methods to predict protein complexes from the interaction data alone without background knowledge of reported complexes.
Prediction of protein complexes
The algorithm we propose for the unsupervised identification of protein complexes implements the following steps.
Purification experiments are combined by pooling them.
Interaction confidences are determined by first identifying preliminary
complexes for bootstrap samples from the set of purifications.
The resulting confidence network is then clustered with the MCL algorithm (http://micans.org/mcl/)
and proteins shared between complexes are identified in a post-processing step.
The last two steps are also used to determine the preliminary complexes for the bootstrap samples
Bootstrapping the Interactome: Unsupervised Identification of Protein Complexes in Yeast
Proceedings of the 12th Annual International Conference on Research in Computational Molecular Biology, RECOMB 2008, Singapore, March 30 - April 2, 2008, Lecture Notes in Computer Science, vol 4955, pp. 3-16, Springer, 2008.
Identifying the topology of protein complexes
We developed an approach for identifying the direct physical interactions and the subcomponent structure of protein complexes predicted from affinity purification assays. We show that the interactions identified with this approach are more accurate in predicting experimentally derived physical interactions than baseline approaches and resolve more satisfactorily the subcomponent structure of the complexes.
ProCope - Protein complex prediction and evaluation
ProCope is a Java software suite for the prediction and evaluation of protein complexes from affinity purification experiment which integrates the major methods for calculating interaction propensities and predicting protein complexes published over the last years. Methods can be accessed via a graphical user interface, command line tools and a Java API.
As a significant part of biological knowledge on protein interactions is available only as free text,
we focus also on the extraction of these interactions from literature with textming methods.
Details can be found at the
Analysis of miRNA regulated TFs
We develop the method MIRTFnet to determine the experimental conditions where certain regulators like miRNAs become active and how they regulate the transcriptome via cascades of other miRNAs, transcription factors (TFs) or kinases.
Details can be found at the