Journal of Bioinformatics and Computational Biology

Microarray-based characterization of tissues, cellular and disease states, and environmental condition and treatment responses provides genome-wide snapshots containing large amounts of invaluable information. However, the lack of inherent structure within the data and strong noise make extracting and interpreting this information and formulating and prioritizing domain relevant hypotheses difficult tasks. Integration with different types of biological data is required to place the expression measurements into a biologically meaningful context. A few approaches in microarray data interpretation are discussed with the emphasis on the use of molecular network information. Statistical procedures are demonstrated that superimpose expression data onto the transcription regulation network mined from scientific literature and aim at selecting transcription regulators with significant patterns of expression changes downstream. Tests are suggested that take into account network topology and signs of transcription regulation effects. The approaches are illustrated using two different expression datasets, the performance is compared, and biological relevance of the predictions is discussed.

In this theoretical paper, we focus to the usage of formal languages and define an extended Backus-System which will allow an adequate representation of molecular data. Furthermore, based on this new formalization we try to define the "complexity of organisms". Our results show that this formalization is useful for the syntactical specification of the genome interpreted as a formal language.

Mechanistic models for transcriptional regulation are derived using the methods of equilibrium statistical mechanics, to model equilibrating processes that occur at a fast time scale. These processes regulate slower changes in the synthesis and expression of transcription factors that feed back and cooperatively regulate transcription, forming a gene regulation network (GRN). We rederive and extend two previous quasi-equilibrium models of transcriptional regulation, and demonstrate circumstances under which they can be approximated at each transcription complex by feed-forward artificial neural network (ANN) models. A single-level mechanistic model can be approximated by a successfully applied phenomenological model of GRNs which is based on single-layer analog-valued ANNs. A two-level hierarchical mechanistic model, with separate activation states for modules and for the whole transcription complex, can be approximated by a two-layer feed-forward ANN in several related ways. The sufficient conditions demonstrated for the ANN approximations correspond biologically to large numbers of binding sites each of which have a small effect. A further extension to the single-level and two-level models allows one-dimensional chains of overlapping and/or energetically interacting binding sites within a module. Partition functions for these models can be constructed from stylized diagrams that indicate energetic and logical interactions between binary-valued state variables. All parameters in the mechanistic models, including the two approximations, can in principle be related to experimentally measurable free energy differences, among other observables.

The model for reception of the concentration gradient of the Hedgehog morphogen has been developed. The mechanism of co-operation of the proteins Patched, Smoothened, and Hedgehog is theoretically analyzed in terms of different versions of interactions within this group of proteins. The parametric stability of the modeled system is considered.

Two novel databases, GenSensor and ConSensor, have been developed. GenSensor accumulates information on the sensitivities of the prokaryotic genes to external stimuli and may facilitate designing of novel genosensors; ConSensor contains data about the structure and efficiency of the available genosensor plasmid constructs. Using these databases, candidate genes for the design of novel multiple functional genosensors were searched, and the Escherichia coli dps gene was chosen as the candidate. The genetic construct derived from its promoter was developed and tested for its sensitivity to various stress agents: hydrogen peroxide (oxidative stress), phenol (protein and membrane damaging), and mitomycin C (DNA damaging). This genosensor was found to be sensitive to all stress conditions applied confirming its ability to serve as multi-functional genosensor. The GenSensor and ConSensor databases are available at .

Development of an in silico cell is an urgent task of systems biology. The core of this cell should consist of mathematical models of intracellular events, including enzymatic reactions and control of gene expression. For example, the minimal model of the E. coli cell should include description of about one thousand enzymatic reactions and regulation of expression of approximately the same number of genes. In many cases detailed mechanisms of molecular processes are not known. In this study, we propose a generalized Hill function method for modeling molecular events. The proposed approach is a method of kinetic data approximation in view of additional data on structure functional features of molecular genetic systems and actually does not demand knowledge of their detailed mechanisms. Generalized Hill function models of an enzymatic reaction catalyzed by the tryptophan-sensitive 3-deoxy-d-arabino-heptulosonate-7-phosphate synthase and the cydAB operon expression regulation are presented.

Recently, a set of highly conserved non-coding elements (CNEs) has been derived from a comparison between the genomes of the puffer fish, Takifugu or Fugu rubripes, and man. In order to facilitate the identification of these conserved elements in silico, we characterize them by a number of statistical features.

We found a pronounced information pattern around CNE borders; although the CNEs themselves are AT rich and have high entropy (complexity), they are flanked by GC-rich regions of low entropy (complexity). We also identified the most abundant motifs within and around of CNEs, and identified those that group around their borders. Like in human promoter regions, the TBP, NF-Y and some other binding motifs are clustered around CNE boundaries, which may suggest a possible transcription regulatory function of CNEs.

Mapping of putative promoters within the entire genome of Escherichia coli (E. coli) by means of pattern-recognition software PlatProm revealed several thousand of sites having high probability to perform promoter function. Along with the expected promoters located upstream of coding sequences, PlatProm identified more than a thousand potential promoters for antisense transcription and several hundred very similar signals within coding sequences having the same direction with the genes. Since recently developed ChIP–chip technology also testified the presence of intragenic RNA polymerase binding sites, such distribution of putative promoters is likely to be a general biological phenomenon reflecting yet undiscovered regulatory events. Here, we provide experimental evidences that two internal promoters are recognized by bacterial RNA polymerase. One of them is located within the hns coding sequence and may initiate synthesis of RNA from the antisense strand. Another one is found within the overlapping genes htgA/yaaW and may control the production of a shortened mRNA or an RNA-product complementary to mRNA of yaaW. Both RNA-products can form secondary structures with free energies of folding close to those of small regulatory RNAs (sRNAs) of the same length. Folding propensity of known sRNAs was further compared with that of antisense RNAs (aRNAs), predicted in E. coli as well as in Salmonella typhimurium (S. typhimurium). Slightly lower stability observed for aRNAs assumes that their structural compactness may be less significant for biological function.

Variable order Markov models and variable order Bayesian trees have been proposed for the recognition of cis-regulatory elements, and it has been demonstrated that they outperform traditional models such as position weight matrices, Markov models, and Bayesian trees for the recognition of binding sites in prokaryotes. Here, we study to which degree variable order models can improve the recognition of eukaryotic cis-regulatory elements. We find that variable order models can improve the recognition of binding sites of all the studied transcription factors. To ease a systematic evaluation of different model combinations based on problem-specific data sets and allow genomic scans of cis-regulatory elements based on fixed and variable order Markov models and Bayesian trees, we provide the VOMBATserver to the public community.

Protein–protein and protein–ligand interactions play a central role in biochemical reactions, and understanding these processes is an important task in different fields of biomedical science and drug discovery. Proteins often work in complex assemblies of several macromolecules and small ligands. The structural and functional description of protein–protein interactions (PPI) is very important for basic-, as well as applied research. The interface areas of protein complexes have unique structure and properties, so PPI represent prospective targets for a new generation of drugs. One of the key targets of PPI inhibitors are oligomeric enzymes. This report shows interactive links between virtual and experimental approaches in a total pipeline "from gene to drug" and using Surface Plasmon Resonance technology for experimentally assessing PPI. Our research is conducted on two oligomeric enzymes — HIV-1 protease (HIVp) (homo-dimer) and bacterial L-asparaginase (homo-tetramer). Using methods of molecular modeling and computational alanine scanning we obtained structural and functional description of PPI in these two enzymes. We also presented a real example of application of integral approach in searching inhibitors of HIVp dimerization — from virtual database mining up to experimental testing of lead compounds.

A mathematical model for suppression of the hepatitis C virus RNA replicon replication in Huh-7 cell culture in the presence of potential drugs was built. There was a good agreement between the experimental and theoretical kinetic data for the decrease in the level of viral RNA in the cell in the presence of the competitive HCV NS3 protease inhibitor. Using the model, we verified the estimates for the efficiency of the effect of potential drugs on replication of viral RNA and viral protein processing. It was demonstrated that the tested drugs are most efficient at the replication step of viral RNA. The efficiency of the combined action of real and putative inhibitors target on the host and viral proteins was also studied. It was found that the action of the inhibitor at low concentrations on the host factors considerably enhances the suppressive effect on viral RNA replication in the presence of even the low affine NS3 protease inhibitors. The developed mathematical model may serve as a tool for the evaluation of the efficiency of potential drugs on the HCV genome.

Structure activity relationships were investigated for membrane-lytic peptides (MLP) Ltc1 and Ltc2a from the latarcin family. The peptides were studied via long-term molecular dynamics (MD) simulations in different membrane environments (detergent micelles, mixed lipid bilayers mimiking eukaryotic and bacterial membranes). The calculated structure of Ltc2a in sodium dodecyl sulfate micelle agrees well with the data obtained by ¹H-NMR spectroscopy. This validates the applied modeling approach. The binding mode of MLPs is governed by several factors: (i) the membrane surface curvature; (ii) the conformational plasticity and hydrophobic organization of the peptide, which depend on the arrangement of charged, non-polar and helix-breaking residues in the amino acid sequence. In contrast to Ltc1, insertion of Ltc2a into model membranes induces significant changes in dynamic behavior of lipids in the contact region. Such a prominent membrane destabilization correlates with high membrane-lytic activity of Ltc2a. In all cases the "membrane response" has a local character and is caused by formation of specific peptide-lipid contacts. Results of MD simulations of Ltc2a in model membranes were used to develop a number of its analogs with predefined activity.

Understanding the functioning of genetic regulatory networks supposes a modeling of biological processes in order to simulate behaviors and to reason on the model. Unfortunately, the modeling task is confronted to incomplete knowledge about the system. To deal with this problem we propose a methodology that uses the qualitative approach developed by Thomas. A symbolic transition system can represent the set of all possible models in a concise and symbolic way. We introduce a new method based on model-checking techniques and symbolic execution to extract constraints on parameters leading to dynamics coherent with known behaviors. Our method allows us to efficiently respond to two kinds of questions: is there any model coherent with a certain hypothetic behavior? Are there behaviors common to all selected models? The first question is illustrated with the example of the mucus production in Pseudomonas aeruginosa while the second one is illustrated with the example of immunity control in bacteriophage lambda.

Development of organisms is a very complex process in which a lot of gene networks of different cell types are integrated. Development of a cellular automaton (Ermentrout and Edelshtein-Keshet, J Theor Biol160:97–133, 1993) that models the morphodynamics of different cell types is the first step in understanding and analysis of the regulatory mechanisms underlying the functioning of developmental gene networks. A model of a cellular automaton has been developed, which simulates the embryonic development of shoot meristem in Arabidopsis thaliana. The model adequately describes the basic stages in development of this organ in wild and mutant types.