Gene Network Model Research Articles

Abstract 4920: Gene network models identify targets for mammary tumor normalization

Abstract The majority of breast cancer therapeutics aim at slowing tumor expansion through selective killing of tumor cells. Given the enormous plasticity of cells (highlighted by recent advances in the iPS field), we instead sought to induce a cell state transition from a mammary tumor cell phenotype to a normal mammary epithelial cell state. We have tackled the critical problem of identifying target genes for the normalization or differentiation of mammary tumor cells by employing a systems biology approach that integrates computational modeling with in vivo and in vitro studies. In vivo gene expression data of mammary tumor progression in the FVB C3(1) SV40Tag transgenic mouse model were collected at multiple time points as the disease progressed from normal gland to DCIS-like lesions to invasive carcinoma. Gene regulatory network models predicted novel transcription factor targets that appeared critical to the rise of the tumor phenotype. These putative gene targets were then assessed in a 3D spheroid assay. RNAi inhibition of two computationally predicted transcription factors reverted mammary tumor cells to a normalized mammary epithelial cell phenotype, with appropriate apical/basal polarization, lumen formation and reduced proliferative index. This integrated strategy has pinpointed transcription factors for RNAi therapeutic targeting in vivo but importantly, the method is not limited to breast cancer and may be useful for many different tumor types. In summary, our major advance has been the demonstration that a gene regulatory network reverse engineering approach can identify genes that are causatively involved in cancer progression and that represent viable therapeutic differentiation targets. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 4920. doi:1538-7445.AM2012-4920

Most networks in Wagner's model are cycling.

In this paper we study a model of gene networks introduced by Andreas Wagner in the 1990s that has been used extensively to study the evolution of mutational robustness. We investigate a range of model features and parameters and evaluate the extent to which they influence the probability that a random gene network will produce a fixed point steady state expression pattern. There are many different types of models used in the literature, (discrete/continuous, sparse/dense, small/large network) and we attempt to put some order into this diversity, motivated by the fact that many properties are qualitatively the same in all the models. Our main result is that random networks in all models give rise to cyclic behavior more often than fixed points. And although periodic orbits seem to dominate network dynamics, they are usually considered unstable and not allowed to survive in previous evolutionary studies. Defining stability as the probability of fixed points, we show that the stability distribution of these networks is highly robust to changes in its parameters. We also find sparser networks to be more stable, which may help to explain why they seem to be favored by evolution. We have unified several disconnected previous studies of this class of models under the framework of stability, in a way that had not been systematically explored before.

Gene network models robust to spatial scaling and noisy input

Many biological systems are inherently noisy, yet demonstrate robustness to perturbations and changes in external influences. Such is the case in the Bicoid–Hunchback (Bcd–Hb) system, which is critical to axis specification in the developing Drosophila embryo. We use this system as motivation to explore the larger problem of how precise patterning can be achieved under imprecise conditions. While evidence suggests Bicoid gradients are uncorrelated with respect to embryo length, downstream genes, such as Hb, are expressed in a precise manner with regard to position along the anterior–posterior (AP)-axis. In addition to precision under variability of embryo length, Hb also exhibits robustness to perturbations to the regulatory network, gene dosage, and temperature. Understanding the reduced variability of patterns in this system is of interest to both experimentalists and theoreticians, lending itself well to the field of mathematical modeling. In this paper, a class of reaction–diffusion models is presented, which produce precise patterns, despite receiving noisy input and other perturbations to the system. An essential property of the network includes the existence of a strong inhibitor for the Hb representative, where the strength of the inhibition is directly related to the amount of variation that can be tolerated. With a higher inhibitory effect, larger perturbations of Bcd can be made with relatively small changes to the location of the Hb boundary. Network topology and interaction strength are the essential properties of the minimal model giving rise to the robust features, and possible interpretations are made with regard to the Bcd–Hb system.

Bayesian Joint Modeling of Multiple Gene Networks and Diverse Genomic Data to Identify Target Genes of a Transcription Factor.

We consider integrative modeling of multiple gene networks and diverse genomic data, including protein-DNA binding, gene expression and DNA sequence data, to accurately identify the regulatory target genes of a transcription factor (TF). Rather than treating all the genes equally and independently a priori in existing joint modeling approaches, we incorporate the biological prior knowledge that neighboring genes on a gene network tend to be (or not to be) regulated together by a TF. A key contribution of our work is that, to maximize the use of all existing biological knowledge, we allow incorporation of multiple gene networks into joint modeling of genomic data by introducing a mixture model based on the use of multiple Markov random fields (MRFs). Another important contribution of our work is to allow different genomic data to be correlated and to examine the validity and effect of the independence assumption as adopted in existing methods. Due to a fully Bayesian approach, inference about model parameters can be carried out based on MCMC samples. Application to an E. coli data set, together with simulation studies, demonstrates the utility and statistical efficiency gains with the proposed joint model.

Adaptive Models for Gene Networks

Biological systems are often treated as time-invariant by computational models that use fixed parameter values. In this study, we demonstrate that the behavior of the p53-MDM2 gene network in individual cells can be tracked using adaptive filtering algorithms and the resulting time-variant models can approximate experimental measurements more accurately than time-invariant models. Adaptive models with time-variant parameters can help reduce modeling complexity and can more realistically represent biological systems.

Analogues of the Smale and Hirsch theorems for cooperative Boolean and other discrete systems

Discrete dynamical systems defined on the state space have been used in multiple applications, most recently for the modelling of gene and protein networks. In this paper, we study to what extent well-known theorems by Smale and Hirsch, which form part of the theory of (continuous) monotone dynamical systems, generalize or fail to do so in the discrete case. We show that arbitrary m-dimensional systems cannot necessarily be embedded into n-dimensional cooperative systems for , as in the Smale theorem for the continuous case, but we show that this is possible for as long as p is sufficiently large. We also prove that strict cooperativity, a natural weakening of the notion of strong cooperativity, implies non-trivial bounds on the lengths of periodic orbits in discrete systems and imposes a condition akin to Lyapunov stability on all attractors. Finally, we explore several natural candidates for definitions of irreducibility of a discrete system. While some of these notions imply the strict cooperativity of a given cooperative system and impose even tighter bounds on the lengths of periodic orbits than strict cooperativity alone, other plausible definitions allow the existence of exponentially long periodic orbits.

BioNetwork Bench: Database and Software for Storage, Query, and Analysis of Gene and Protein Networks

Gene and protein networks offer a powerful approach for integration of the disparate yet complimentary types of data that result from high-throughput analyses. Although many tools and databases are currently available for accessing such data, they are left unutilized by bench scientists as they generally lack features for effective analysis and integration of both public and private datasets and do not offer an intuitive interface for use by scientists with limited computational expertise. We describe BioNetwork Bench, an open source, user-friendly suite of database and software tools for constructing, querying, and analyzing gene and protein network models. It enables biologists to analyze public as well as private gene expression; interactively query gene expression datasets; integrate data from multiple networks; store and selectively share the data and results. Finally, we describe an application of BioNetwork Bench to the assembly and iterative expansion of a gene network that controls the differentiation of retinal progenitor cells into rod photoreceptors. The tool is available from http://bionetworkbench.sourceforge.net/BackgroundThe emergence of high-throughput technologies has allowed many biological investigators to collect a great deal of information about the behavior of genes and gene products over time or during a particular disease state. Gene and protein networks offer a powerful approach for integration of the disparate yet complimentary types of data that result from such high-throughput analyses. There are a growing number of public databases, as well as tools for visualization and analysis of networks. However, such databases and tools have yet to be widely utilized by bench scientists, as they generally lack features for effective analysis and integration of both public and private datasets and do not offer an intuitive interface for use by biological scientists with limited computational expertise.ResultsWe describe BioNetwork Bench, an open source, user-friendly suite of database and software tools for constructing, querying, and analyzing gene and protein network models. BioNetwork Bench currently supports a broad class of gene and protein network models (eg, weighted and un-weighted, undirected graphs, multi-graphs). It enables biologists to analyze public as well as private gene expression, macromolecular interaction and annotation data; interactively query gene expression datasets; integrate data from multiple networks; query multiple networks for interactions of interest; store and selectively share the data as well as results of analyses. BioNetwork Bench is implemented as a plug-in for, and hence is fully interoperable with, Cytoscape, a popular open-source software suite for visualizing macromolecular interaction networks. Finally, we describe an application of BioNetwork Bench to the problem of assembly and iterative expansion of a gene network that controls the differentiation of retinal progenitor cells into rod photoreceptors.ConclusionsBioNetwork Bench provides a suite of open source software for construction, querying, and selective sharing of gene and protein networks. Although initially aimed at a community of biologists interested in retinal development, the tool can be adapted easily to work with other biological systems simply by populating the associated database with the relevant datasets.

HEB in the Spotlight: Transcriptional Regulation of T-Cell Specification, Commitment, and Developmental Plasticity

The development of T cells from multipotent progenitors in the thymus occurs by cascades of interactions between signaling molecules and transcription factors, resulting in the loss of alternative lineage potential and the acquisition of the T-cell functional identity. These processes require Notch signaling and the activity of GATA3, TCF1, Bcl11b, and the E-proteins HEB and E2A. We have shown that HEB factors are required to inhibit the thymic NK cell fate and that HEBAlt allows the passage of T-cell precursors from the DN to DP stage but is insufficient for suppression of the NK cell lineage choice. HEB factors are also required to enforce the death of cells that have not rearranged their TCR genes. The synergistic interactions between Notch1, HEBAlt, HEBCan, GATA3, and TCF1 are presented in a gene network model, and the influence of thymic stromal architecture on lineage choice in the thymus is discussed.

Classifying Lung Cancer Recurrence Time Using Novel Ensemble Method with Gene Network based Input Models

An accurate prognostic model of a cancer patient after treatment can be useful in deciding the next course of treatment or efficacy of said treatment. Gene expression microarray data has been used to predict survival times [1], or to classify the patient as having a good/poor prognosis [2] by predicting whether the patient belongs to the class that will have a recurrence of cancer before or after a certain period, typically 3 or 5 years. Microarrays typically contain thousands of gene expression probes and a typical study may only contain a few hundred patients or less. Typical regression techniques will fail to generalize, suffering from the ‘Curse of Dimensionality’, resulting in an over-fitted model that performs very well on the training data, and very poorly or validation data. Various feature selection/reduction methods have been used to reduce the dimensionality of the data and improve or facilitate a solution [3]. Gene expression is known to be modulated by the expression of other genes, forming a so-called gene network or pathway. Furthermore, several networks may affect the aggressiveness of the cancer simultaneously [4]. While past studies have selected features based on statistical methods alone [5] or have simply included ‘known cancer genes’, none to our knowledge have used classification models based on ensembles of models based on multiple known gene networks. Based on the data presented in Shedden, et. al. [6], this study uses a General Regression Neural Network (GRNN) Oracle ensemble that combines several Partial least squares (PLS) models trained to predict recurrence times from 12 different gene networks. We hypothesize that it is possible to correctly classify recurrence by combining the results based on the gene network models.

The Tenth Asia Pacific Bioinformatics Conference (APBC 2012)

The Tenth Asia Pacific Bioinformatics Conference (APBC 2012)

Transcriptomic analysis reveals pH-responsive antioxidant gene networks

Reactive oxygen species (ROS) are produced in different physiological conditions. In response to ROS imbalance cells activate oxidative stress defenses, which include more than 60 antioxidant genes. It has been suggested that gene products associated with ROS detoxification can work coordinately, acting as an antioxidant-defense network. However, the functional overlap among oxidative stress defenses and other related cell functions makes difficult the characterization of this network. We previously described a network-based model to characterize the interactions existing among different antioxidant gene products and their substrates. Here, we test whether this network-based model of human antioxidant genes can respond to different physiological conditions. We used a systems biology approach applied to the analysis of two independent gene expression datasets: transcriptomes from HeLa cells and primary astrocytes maintained under hypoxic conditions and transcriptomes from SKGT4 cells exposed to low pH environment. We found that the proposed gene network model responds selectively to both hypoxia and acidosis. We anticipate that this antioxidant gene network model can be helpful to describe stress-responsive expression profiles in different cell types.

Haploid evolutionary constructor: New features and further challenges

In this paper we consider the recent advances in methodology for modeling of prokaryotic communities evolution and new features of the software package "Haploid evolutionary constructor" (http://evol-constructor.bionet.nsc.ru). We show the principles of building complex computer models in our software tool. These models describe several levels of biological organization: genetic, metabolic, population, ecological. New features of the haploid evolutionary constructor include the modeling of gene networks and phage infections.

Affymetrix GeneChip microarray preprocessing for multivariate analyses.

Affymetrix GeneChip microarrays are the most widely used high-throughput technology to measure gene expression, and a wide variety of preprocessing methods have been developed to transform probe intensities reported by a microarray scanner into gene expression estimates. There have been numerous comparisons of these preprocessing methods, focusing on the most common analyses-detection of differential expression and gene or sample clustering. Recently, more complex multivariate analyses, such as gene co-expression, differential co-expression, gene set analysis and network modeling, are becoming more common; however, the same preprocessing methods are typically applied. In this article, we examine the effect of preprocessing methods on some of these multivariate analyses and provide guidance to the user as to which methods are most appropriate.

Evolutionary Synthesis of Stochastic Gene Network Models Using Feature-based Search Spaces

A feature-based fitness function is applied in a genetic programming system to synthesize stochastic gene regulatory network models whose behaviour is defined by a time course of protein expression levels. Typically, when targeting time series data, the fitness function is based on a sum-of-errors involving the values of the fluctuating signal. While this approach is successful in many instances, its performance can deteriorate in the presence of noise and/or stochastic behaviour. This paper explores a fitness measure determined from a set of statistical features characterizing the time series’ sequence of values, rather than the actual values themselves. Through a series of experiments involving modular gene regulatory network models based on the stochastic π-calculus, it is shown to successfully target oscillating and non-oscillating signals. This practical and versatile fitness function offers an alternate approach, worthy of consideration for use in algorithms that evaluate noisy or stochastic behaviour.

Automatic detection of changes in the dynamics of delayed stochastic gene networks and in vivo production of RNA molecules in Escherichia coli

Production and degradation of RNA and proteins are stochastic processes, difficulting the distinction between spurious fluctuations in their numbers and changes in the dynamics of a genetic circuit. An accurate method of change detection is key to analyze plasticity and robustness of stochastic genetic circuits. We use automatic change point detection methods to detect non-spurious changes in the dynamics of delayed stochastic models of gene networks at run time. We test the methods in detecting changes in mean and noise of protein numbers, and in the switching frequency of a genetic switch. We also detect changes, following genes' silencing, in the dynamics of a model of the core gene regulatory network of Saccharomyces cerevisiae with 328 genes. Finally, from images, we determine when RNA molecules tagged with fluorescent proteins are first produced in Escherichia coli. Provided prior knowledge on the time scale of the changes, the methods detect them accurately and are robust to fluctuations in protein and RNA levels. Simulator: www.cs.tut.fi/~sanchesr/SGN/SGNSim.html andre.ribeiro@tut.fi Supplementary data are available at Bioinformatics online.

Order-preserving principles underlying genotype-phenotype maps ensure high additive proportions of genetic variance

In quantitative genetics, the degree of resemblance between parents and offspring is described in terms of the additive variance (V(A)) relative to genetic (V(G)) and phenotypic (V(P)) variance. For populations with extreme allele frequencies, high V(A)/V(G) can be explained without considering properties of the genotype-phenotype (GP) map. We show that randomly generated GP maps in populations with intermediate allele frequencies generate far lower V(A)/V(G) values than empirically observed. The main reason is that order-breaking behaviour is ubiquitous in random GP maps. Rearrangement of genotypic values to introduce order-preservation for one or more loci causes a dramatic increase in V(A)/V(G). This suggests the existence of order-preserving design principles in the regulatory machinery underlying GP maps. We illustrate this feature by showing how the ubiquitously observed monotonicity of dose-response relationships gives much higher V(A)/V(G) values than a unimodal dose-response relationship in simple gene network models.

Integrating Rare-Variant Testing, Function Prediction, and Gene Network in Composite Resequencing-Based Genome-Wide Association Studies (CR-GWAS)

High-density array-based genome-wide association studies (GWAS) are complemented by exome sequencing and whole-genome resequencing-based association studies. Here we present a composite resequencing-based genome-wide association study (CR-GWAS) strategy that systematically exploits collective biological information and analytical tools for a robust analysis. We showcased the utility of this strategy by using Arabidopsis (Arabidopsis thaliana) resequencing data. Bioinformatic predictions of biological function alteration at each locus were integrated into the process of association testing of both common and rare variants for complex traits with a suite of statistics. Significant signals were then filtered with a priori candidate loci generated from genome database and gene network models to obtain a posteriori candidate loci. A probabilistic gene network (AraNet) that interrogates network neighborhoods of genes was then used to expand the filtering power to examine the significant testing signals. Using this strategy, we confirmed the known true positives and identified several new promising associations. Promising genes (AP1, FCA, FRI, FLC, FLM, SPL5, FY, and DCL2) were shown to control for flowering time through either common variants or rare variants within a diverse set of Arabidopsis accessions. Although many of these candidate genes were cloned earlier with mutational studies, identifying their allele variation contribution to overall phenotypic variation among diverse natural accessions is critical. Our rare allele testing established a greater number of connections than previous analyses in which this issue was not addressed. More importantly, our results demonstrated the potential of integrating various biological, statistical, and bioinformatic tools into complex trait dissection.

Weighted lasso in graphical Gaussian modeling for large gene network estimation based on microarray data.

We propose a statistical method based on graphical Gaussian models for estimating large gene networks from DNA microarray data. In estimating large gene networks, the number of genes is larger than the number of samples, we need to consider some restrictions for model building. We propose weighted lasso estimation for the graphical Gaussian models as a model of large gene networks. In the proposed method, the structural learning for gene networks is equivalent to the selection of the regularization parameters included in the weighted lasso estimation. We investigate this problem from a Bayes approach and derive an empirical Bayesian information criterion for choosing them. Unlike Bayesian network approach, our method can find the optimal network structure and does not require to use heuristic structural learning algorithm. We conduct Monte Carlo simulation to show the effectiveness of the proposed method. We also analyze Arabidopsis thaliana microarray data and estimate gene networks.

Robust graphical modeling of gene networks using classical and alternative t-distributions

Graphical Gaussian models have proven to be useful tools for exploring network structures based on multivariate data. Applications to studies of gene expression have generated substantial interest in these models, and resulting recent progress includes the development of fitting methodology involving penalization of the likelihood function. In this paper we advocate the use of multivariate t-distributions for more robust inference of graphs. In particular, we demonstrate that penalized likelihood inference combined with an application of the EM algorithm provides a computationally efficient approach to model selection in the t-distribution case. We consider two versions of multivariate t-distributions, one of which requires the use of approximation techniques. For this distribution, we describe a Markov chain Monte Carlo EM algorithm based on a Gibbs sampler as well as a simple variational approximation that makes the resulting method feasible in large problems.

An Adaptive GA--PSO Approach with Gene Clustering to Infer S-system Models of Gene Regulatory Networks

The construction of gene regulatory networks from expression data is one of the most important issues in systems biology research. However, building such networks is a tedious task, especially when both the number of genes and the complexity of gene regulation increase. In this work, we adopt the S-system model to represent the gene network and establish a methodology to infer the model. Our work mainly includes an adaptive genetic algorithm-particle swarm optimization hybrid method to infer appropriate network parameters, and a gene clustering method to decompose a large network into several smaller networks for dimension reduction. To validate the proposed methods, different series of experiments have been conducted and the results show that the proposed methods can be used to infer S-system models of gene networks efficiently and successfully.