A scale-free structure prior for graphical models with applications in functional genomics

The problem of reconstructing large-scale, gene regulatory networks from gene expression data has garnered considerable attention in bioinformatics over the past decade with the graphical modeling paradigm having emerged as a popular framework for inference. Analysis in a full Bayesian setting is contingent upon the assignment of a so-called structure prior-a probability distribution on networks, encoding a priori biological knowledge either in the form of supplemental data or high-level topological features. A key topological consideration is that a wide range of cellular networks are approximately scale-free, meaning that the fraction, , of nodes in a network with degree is roughly described by a power-law with exponent between and . The standard practice, however, is to utilize a random structure prior, which favors networks with binomially distributed degree distributions. In this paper, we introduce a scale-free structure prior for graphical models based on the formula for the probability of a network under a simple scale-free network model. Unlike the random structure prior, its scale-free counterpart requires a node labeling as a parameter. In order to use this prior for large-scale network inference, we design a novel Metropolis-Hastings sampler for graphical models that includes a node labeling as a state space variable. In a simulation study, we demonstrate that the scale-free structure prior outperforms the random structure prior at recovering scale-free networks while at the same time retains the ability to recover random networks. We then estimate a gene association network from gene expression data taken from a breast cancer tumor study, showing that scale-free structure prior recovers hubs, including the previously unknown hub SLC39A6, which is a zinc transporter that has been implicated with the spread of breast cancer to the lymph nodes. Our analysis of the breast cancer expression data underscores the value of the scale-free structure prior as an instrument to aid in the identification of candidate hub genes with the potential to direct the hypotheses of molecular biologists, and thus drive future experiments.     

Gene regulatory network reconstruction by Bayesian integration of prior knowledge and/or different experimental conditions

There have been various attempts to improve the reconstruction of gene regulatory networks from microarray data by the systematic integration of biological prior knowledge. Our approach is based on pioneering work by Imoto et al. where the prior knowledge is expressed in terms of energy functions, from which a prior distribution over network structures is obtained in the form of a Gibbs distribution. The hyperparameters of this distribution represent the weights associated with the prior knowledge relative to the data. We have derived and tested a Markov chain Monte Carlo (MCMC) scheme for sampling networks and hyperparameters simultaneously from the posterior distribution, thereby automatically learning how to trade off information from the prior knowledge and the data. We have extended this approach to a Bayesian coupling scheme for learning gene regulatory networks from a combination of related data sets, which were obtained under different experimental conditions and are therefore potentially associated with different active subpathways. The proposed coupling scheme is a compromise between (1) learning networks from the different subsets separately, whereby no information between the different experiments is shared; and (2) learning networks from a monolithic fusion of the individual data sets, which does not provide any mechanism for uncovering differences between the network structures associated with the different experimental conditions. We have assessed the viability of all proposed methods on data related to the Raf signaling pathway, generated both synthetically and in cytometry experiments.     

Modelling the evolution of genetic regulatory networks

An evolutionary model of genetic regulatory networks is developed, based on a model of network encoding and dynamics called the Artificial Genome (AG). This model derives a number of specific genes and their interactions from a string of (initially random) bases in an idealized manner analogous to that employed by natural DNA. The gene expression dynamics are determined by updating the gene network as if it were a simple Boolean network. The generic behaviour of the AG model is investigated in detail. In particular, we explore the characteristic network topologies generated by the model, their dynamical behaviours, and the typical variance of network connectivities and network structures. These properties are demonstrated to agree with a probabilistic analysis of the model, and the typical network structures generated by the model are shown to lie between those of random networks and scale-free networks in terms of their degree distribution. Evolutionary processes are simulated using a genetic algorithm, with selection acting on a range of properties from gene number and degree of connectivity through periodic behaviour to specific patterns of gene expression. The evolvability of increasingly complex patterns of gene expression is examined in detail. When a degree of redundancy is introduced, the average number of generations required to evolve given targets is reduced, but limits on evolution of complex gene expression patterns remain. In addition, cyclic gene expression patterns with periods that are multiples of shorter expression patterns are shown to be inherently easier to evolve than others. Constraints imposed by the template-matching nature of the AG model generate similar biases towards such expression patterns in networks in initial populations, in addition to the somewhat scale-free nature of these networks. The significance of these results on current understanding of biological evolution is discussed.     

Wrestling with pleiotropy: genomic and topological analysis of the yeast gene expression network

The vast majority (>95%) of single-gene mutations in yeast affect not only the expression of the mutant gene, but also the expression of many other genes. These data suggest the presence of a previously uncharacterized "gene expression network"--a set of interactions between genes which dictate gene expression in the native cell environment. Here, we quantitatively analyze the gene expression network revealed by microarray expression data from 273 different yeast gene deletion mutants.(1) We find that gene expression interactions form a robust, error-tolerant "scale-free" network, similar to metabolic pathways(2) and artificial networks such as power grids and the internet.(3-5) Because the connectivity between genes in the gene expression network is unevenly distributed, a scale-free organization helps make organisms resistant to the deleterious effects of mutation, and is thus highly adaptive. The existence of a gene expression network poses practical considerations for the study of gene function, since most mutant phenotypes are the result of changes in the expression of many genes. Using principles of scale-free network topology, we propose that fragmenting the gene expression network via "genome-engineering" may be a viable and practical approach to isolating gene function.     

Reverse engineering of regulatory networks: simulation studies on a genetic algorithm approach for ranking hypotheses

Reverse engineering algorithms (REAs) aim at using gene expression data to reconstruct interactions in regulatory genetic networks. This may help to understand the basis of gene regulation, the core task of functional genomics. Collecting data for a number of environmental conditions is necessary to reengineer even the smallest regulatory networks with reasonable confidence. We systematically tested the requirements for the experimental design necessary for ranking alternative hypotheses about the structure of a given regulatory network. A genetic algorithm (GA) was used to explore the parameter space of a multistage discrete genetic network model with fixed connectivity and number of states per node. Our results show that it is not necessary to determine all parameters of the genetic network in order to rank hypotheses. The ranking process is easier the more experimental environmental conditions are used for the data set. During the ranking, the number of fixed parameters increases with the number of environmental conditions, while some errors in the hypothetical network structure may pass undetected, due to a maintained dynamical behaviour.     

Integrating external biological knowledge in the construction of regulatory networks from time-series expression data

ABSTRACT: BACKGROUND: Inference about regulatory networks from high-throughput genomics data is of great interest in systems biology. We present a Bayesian approach to infer gene regulatory networks from time series expression data by integrating various types of biological knowledge. RESULTS: We formulate network construction as a series of variable selection problems and use linear regression to model the data. Our method summarizes additional data sources with an informative prior probability distribution over candidate regression models. We extend the Bayesian model averaging (BMA) variable selection method to select regulators in the regression framework. We summarize the external biological knowledge by an informative prior probability distribution over the candidate regression models. CONCLUSIONS: We demonstrate our method on simulated data and a set of time-series microarray experiments measuring the effect of a drug perturbation on gene expression levels, and show that it outperforms leading regression-based methods in the literature.     

Do scale-free regulatory networks allow more expression than random ones?

In this paper, we compile the network of software packages with regulatory interactions (dependences and conflicts) from Debian GNU/Linux operating system and use it as an analogy for a gene regulatory network. Using a trace-back algorithm we assemble networks from the pool of packages with both scale-free (real data) and exponential (null model) topologies. We record the maximum number of packages that can be functionally installed in the system (i.e., the active network size). We show that scale-free regulatory networks allow a larger active network size than random ones. This result might have implications for the number of expressed genes at steady state. Small genomes with scale-free regulatory topologies could allow much more expression than large genomes with exponential topologies. This may have implications for the dynamics, robustness and evolution of genomes.     

Emergence of scale-free distribution in protein–protein interaction networks based on random selection of interacting domain pairs

Recent analyses of biological and artificial networks have revealed a common network architecture, called scale-free topology. The origin of the scale-free topology has been explained by using growth and preferential attachment mechanisms. In a cell, proteins are the most important carriers of function, and are composed of domains as elemental units responsible for the physical interaction between protein pairs. Here, we propose a model for protein–protein interaction networks that reveals the emergence of two possible topologies. We show that depending on the number of randomly selected interacting domain pairs, the connectivity distribution follows either a scale-free distribution, even in the absence of the preferential attachment, or a normal distribution. This new approach only requires an evolutionary model of proteins (nodes) but not for the interactions (edges). The edges are added by means of random interaction of domain pairs. As a result, this model offers a new mechanistic explanation for understanding complex networks with a direct biological interpretation because only protein structures and their functions evolved through genetic modifications of amino acid sequences. These findings are supported by numerical simulations as well as experimental data.     

Gene network inference from incomplete expression data: transcriptional control of hematopoietic commitment

MOTIVATION: The topology and function of gene regulation networks are commonly inferred from time series of gene expression levels in cell populations. This strategy is usually invalid if the gene expression in different cells of the population is not synchronous. A promising, though technically more demanding alternative is therefore to measure the gene expression levels in single cells individually. The inference of a gene regulation network requires knowledge of the gene expression levels at successive time points, at least before and after a network transition. However, owing to experimental limitations a complete determination of the precursor state is not possible. RESULTS: We investigate a strategy for the inference of gene regulatory networks from incomplete expression data based on dynamic Bayesian networks. This permits prediction of the number of experiments necessary for network inference depending on parameters including noise in the data, prior knowledge and limited attainability of initial states. Our strategy combines a gradual 'Partial Learning' approach based solely on true experimental observations for the network topology with expectation maximization for the network parameters. We illustrate our strategy by extensive computer simulations in a high-dimensional parameter space in a simulated single-cell-based example of hematopoietic stem cell commitment and in random networks of different sizes. We find that the feasibility of network inferences increases significantly with the experimental ability to force the system into different initial network states, with prior knowledge and with noise reduction. AVAILABILITY: Source code is available under: www.izbi.uni-leipzig.de/services/NetwPartLearn.html SUPPLEMENTARY INFORMATION: Supplementary Data are available at Bioinformatics online.     

Integration of Steady-State and Temporal Gene Expression Data for the Inference of Gene Regulatory Networks

We develop a new regression algorithm, cMIKANA, for inference of gene regulatory networks from combinations of steady-state and time-series gene expression data. Using simulated gene expression datasets to assess the accuracy of reconstructing gene regulatory networks, we show that steady-state and time-series data sets can successfully be combined to identify gene regulatory interactions using the new algorithm. Inferring gene networks from combined data sets was found to be advantageous when using noisy measurements collected with either lower sampling rates or a limited number of experimental replicates. We illustrate our method by applying it to a microarray gene expression dataset from human umbilical vein endothelial cells (HUVECs) which combines time series data from treatment with growth factor TNF and steady state data from siRNA knockdown treatments. Our results suggest that the combination of steady-state and time-series datasets may provide better prediction of RNA-to-RNA interactions, and may also reveal biological features that cannot be identified from dynamic or steady state information alone. Finally, we consider the experimental design of genomics experiments for gene regulatory network inference and show that network inference can be improved by incorporating steady-state measurements with time-series data.     

Algebraic connectivity may explain the evolution of gene regulatory networks

Gene expression is a result of the interplay between the structure, type, kinetics, and specificity of gene regulatory interactions, whose diversity gives rise to the variety of life forms. As the dynamic behavior of gene regulatory networks depends on their structure, here we attempt to determine structural reasons which, despite the similarities in global network properties, may explain the large differences in organismal complexity. We demonstrate that the algebraic connectivity, the smallest non-trivial eigenvalue of the Laplacian, of the directed gene regulatory networks decreases with the increase of organismal complexity, and may therefore explain the difference between the variety of analyzed regulatory networks. In addition, our results point out that, for the species considered in this study, evolution favours decreasing concentration of strategically positioned feed forward loops, so that the network as a whole can increase the specificity towards changing environments. Moreover, contrary to the existing results, we show that the average degree, the length of the longest cascade, and the average cascade length of gene regulatory networks cannot recover the evolutionary relationships between organisms. Whereas the dynamical properties of special subnetworks are relatively well understood, there is still limited knowledge about the evolutionary reasons for the already identified design principles pertaining to these special subnetworks, underlying the global quantitative features of gene regulatory networks of different organisms. The behavior of the algebraic connectivity, which we show valid on gene regulatory networks extracted from curated databases, can serve as an additional evolutionary principle of organism-specific regulatory networks.