On entropy and information in gene interaction networks.

MOTIVATION: Modern biological experiments often produce candidate lists of genes presumably related to the studied phenotype. One can ask if the gene list as a whole makes sense in the context of existing knowledge: Are the genes in the list reasonably related to each other or do they look like a random assembly? There are also situations when one wants to know if two or more gene sets are closely related. Gene enrichment tests based on counting the number of genes two sets have in common are adequate if we presume that two genes are related only when they are in fact identical. If by related we mean well connected in the interaction network space, we need a new measure of relatedness for gene sets. RESULTS: We derive entropy, interaction information and mutual information for gene sets on interaction networks, starting from a simple phenomenological model of a living cell. Formally, the model describes a set of interacting linear harmonic oscillators in thermal equilibrium. Because the energy function is a quadratic form of the degrees of freedom, entropy and all other derived information quantities can be calculated exactly. We apply these concepts to estimate the probability that genes from several independent genome-wide association studies are not mutually informative; to estimate the probability that two disjoint canonical metabolic pathways are not mutually informative; and to infer relationships among human diseases based on their gene signatures. We show that the present approach is able to predict observationally validated relationships not detectable by gene enrichment methods. The converse is also true; the two methods are therefore complementary. AVAILABILITY AND IMPLEMENTATION: The functions defined in this paper are available in an R package, gsia, available for download at https://github.com/ucsd-ccbb/gsia.

Transcriptional repression of IFNß1 by ATF2 confers melanoma resistance to therapy.

The resistance of melanoma to current treatment modalities represents a major obstacle for durable therapeutic response, and thus the elucidation of mechanisms of resistance is urgently needed. The crucial functions of activating transcription factor-2 (ATF2) in the development and therapeutic resistance of melanoma have been previously reported, although the precise underlying mechanisms remain unclear. Here, we report a protein kinase C-ɛ (PKCɛ)- and ATF2-mediated mechanism that facilitates resistance by transcriptionally repressing the expression of interferon-ß1 (IFNß1) and downstream type-I IFN signaling that is otherwise induced upon exposure to chemotherapy. Treatment of early-stage melanomas expressing low levels of PKCɛ with chemotherapies relieves ATF2-mediated transcriptional repression of IFNß1, resulting in impaired S-phase progression, a senescence-like phenotype and increased cell death. This response is lost in late-stage metastatic melanomas expressing high levels of PKCɛ. Notably, nuclear ATF2 and low expression of IFNß1 in melanoma tumor samples correlates with poor patient responsiveness to biochemotherapy or neoadjuvant IFN-α2a. Conversely, cytosolic ATF2 and induction of IFNß1 coincides with therapeutic responsiveness. Collectively, we identify an IFNß1-dependent, cell-autonomous mechanism that contributes to the therapeutic resistance of melanoma via the PKCɛ-ATF2 regulatory axis.

Network genomics.

Network genomics is an emerging area of bioengineering which models the influence of genes (hence, genomics) in the context of a larger biomolecular system or network. A biomolecular network is a comprehensive collection of molecules and molecular interactions that regulate cellular function. Molecular interactions include physical binding events between proteins and proteins, proteins and DNA, or proteins and drugs, as well as genetic relationships dictating how genes combine to cause particular phenotypes. Thinking about biological systems as networks goes hand-in-hand with our ability to experimentally measure and define biomolecular interactions at large scale. Once we have catalogued all of the interactions present in a network, we may begin to ask questions such as: How many different molecules are bound by a typical protein? What is the topological structure of the network? How are signals transmitted through the network in response to internal and external events? Which parts of the network are evolutionarily conserved across species, and which parts differ? Perhaps most importantly, we can begin to use the interaction network as a storehouse of information from which to extract and construct computer-based models of cellular processes and disease.

A new approach to decoding life: systems biology.

Systems biology studies biological systems by systematically perturbing them (biologically, genetically, or chemically); monitoring the gene, protein, and informational pathway responses; integrating these data; and ultimately, formulating mathematical models that describe the structure of the system and its response to individual perturbations. The emergence of systems biology is described, as are several examples of specific systems approaches.

Integrated genomic and proteomic analyses of a systematically perturbed metabolic network.

We demonstrate an integrated approach to build, test, and refine a model of a cellular pathway, in which perturbations to critical pathway components are analyzed using DNA microarrays, quantitative proteomics, and databases of known physical interactions. Using this approach, we identify 997 messenger RNAs responding to 20 systematic perturbations of the yeast galactose-utilization pathway, provide evidence that approximately 15 of 289 detected proteins are regulated posttranscriptionally, and identify explicit physical interactions governing the cellular response to each perturbation. We refine the model through further iterations of perturbation and global measurements, suggesting hypotheses about the regulation of galactose utilization and physical interactions between this and a variety of other metabolic pathways.

Discovery of regulatory interactions through perturbation: inference and experimental design.

We present two methods to be used interactively to infer a genetic network from gene expression measurements. The predictor method determines the set of Boolean networks consistent with an observed set of steady-state gene expression profiles, each generated from a different perturbation to the genetic network. The chooser method uses an entropy-based approach to propose an additional perturbation experiment to discriminate among the set of hypothetical networks determined by the predictor. These methods may be used iteratively and interactively to successively refine the genetic network: at each iteration, the perturbation selected by the chooser is experimentally performed to generate a new gene expression profile, and the predictor is used to derive a refined set of hypothetical gene networks using the cumulative expression data. Performance of the predictor and chooser is evaluated on simulated networks with varying number of genes and number of interactions per gene.

Testing for differentially-expressed genes by maximum-likelihood analysis of microarray data.

Although two-color fluorescent DNA microarrays are now standard equipment in many molecular biology laboratories, methods for identifying differentially expressed genes in microarray data are still evolving. Here, we report a refined test for differentially expressed genes which does not rely on gene expression ratios but directly compares a series of repeated measurements of the two dye intensities for each gene. This test uses a statistical model to describe multiplicative and additive errors influencing an array experiment, where model parameters are estimated from observed intensities for all genes using the method of maximum likelihood. A generalized likelihood ratio test is performed for each gene to determine whether, under the model, these intensities are significantly different. We use this method to identify significant differences in gene expression among yeast cells growing in galactose-stimulating versus non-stimulating conditions and compare our results with current approaches for identifying differentially-expressed genes. The effect of sample size on parameter optimization is also explored, as is the use of the error model to compare the within- and between-slide intensity variation intrinsic to an array experiment.

Negative selection: a method for obtaining low-abundance cDNAs using high-density cDNA clone arrays.

The identification of the entire complement of genes expressed in a cell, tissue, or organism provides a framework for understanding biological properties and establishes a tool set for subsequent functional studies. The large-scale sequencing of randomly selected clones from cDNA libraries has been successfully employed as a method for identifying a large fraction of these expressed genes. However, this approach is limited by the inherent redundancy of cellular transcripts reflecting widely variant levels of gene transcription. As a result, a high percentage of transcript duplications are encountered as the number of sequenced clones accrues. To address this problem, we have developed a negative hybridization selection method that employs the hybridization of complex cDNA probes to high-density arrays of cDNA clones and the subsequent selection of clones with a null or low hybridization signal. This approach was applied to a cDNA library constructed from normal human prostate tissue and resulted in the reduction of highly expressed prostate cDNAs from 6.8 to 0.57% with an overall decline in clone redundancy from 33 to 11%. The selected clones also reflected a more diverse cDNA population, with 89% of the clones representing distinctly different cDNAs compared with 67% of the randomly selected clones. This method compares favorably with cDNA library re-association normalization approaches and offers several distinct advantages, including the flexibility to use previously prepared libraries, and the ability to employ an iterative screening approach for continued accrual of cDNAs representing rare transcripts.

Mining SNPs from EST databases.

There is considerable interest in the discovery and characterization of single nucleotide polymorphisms (SNPs) to enable the analysis of the potential relationships between human genotype and phenotype. Here we present a strategy that permits the rapid discovery of SNPs from publicly available expressed sequence tag (EST) databases. From a set of ESTs derived from 19 different cDNA libraries, we assembled 300,000 distinct sequences and identified 850 mismatches from contiguous EST data sets (candidate SNP sites), without de novo sequencing. Through a polymerase-mediated, single-base, primer extension technique, Genetic Bit Analysis (GBA), we confirmed the presence of a subset of these candidate SNP sites and have estimated the allele frequencies in three human populations with different ethnic origins. Altogether, our approach provides a basis for rapid and efficient regional and genome-wide SNP discovery using data assembled from sequences from different libraries of cDNAs.

Effect of electrode position on outcome of low-energy intracardiac cardioversion of atrial fibrillation.

The aim of this study was to evaluate the new method of low-energy, catheter-based intracardiac cardioversion in patients with chronic atrial fibrillation (AF) and to compare 2 different lead positions. Accordingly, we prospectively studied 80 consecutive patients with chronic AF (9.8 +/- 7.9 months) who were randomly assigned to undergo internal cardioversion either via defibrillation electrodes placed in the right atrium and coronary sinus (coronary sinus group) or via defibrillation electrodes placed in the right atrium and left pulmonary artery (pulmonary artery group). Intracardiac shocks were delivered by an external defibrillator synchronized to the QRS complex. After conversion, all patients were treated orally with sotalol (mean daily dose, 189 +/- 63 mg/day). For conversion to sinus rhythm, the overall mean energy requirement was 5.6 +/- 3.1 J. In the coronary sinus group, cardioversion was achieved in 35 of 38 patients at a mean energy level of 4.1 +/- 2.3 J (range 1.0 to 9.9), and in the pulmonary artery group in 39 of 42 patients with 7.2 +/- 3.1 J (range 2.5 to 14.8). Although there was no difference with regard to success rate, the energy differed significantly between the 2 groups (p < 0.01). Mean lead impedance was 56.4 +/- 7.0 omega and 54.6 +/- 8.5 omega, respectively (p = NS). No serious complications were observed in either lead group. At a mean follow-up of 14.2 +/- 7.0 months, 54% and 56%, respectively, of patients who had been converted successfully remained in sinus rhythm. Thus, low-energy biphasic shocks delivered between the right atrium and coronary sinus or pulmonary artery are equally effective for cardioversion of patients with chronic AF. The energy requirements for conversion from a pulmonary artery electrode position are higher than for the coronary sinus position.