Journal of Bioinformatics and Computational Biology

The complex process of genome-scale metabolic network reconstruction involves semi-automatic reaction inference, analysis, and refinement through curation by human experts. Unfortunately, decisions by experts are hampered by the complexity of the network, which can mask errors in the inferred network. In order to aid an expert in making sense out of the thousands of reactions in the organism's metabolism, we developed a method for knowledge-based generalization that provides a higher-level view of the network, highlighting the particularities and essential structure, while hiding the details. In this study, we show the application of this generalization method to 1,286 metabolic networks of organisms in Path2Models that describe fatty acid metabolism. We compare the generalised networks and show that we successfully highlight the aspects that are important for their curation and comparison.

In this paper, a specific aspect of the prediction problem is considered: high predictive power is understood as a possibility to reproduce correct behavior of model solutions at predefined values of a subset of parameters. The problem is discussed in the context of a specific mathematical model, the gene circuit model for segmentation gap gene system in early Drosophila development. A shortcoming of the model is that it cannot be used for predicting the system behavior in mutants when fitted to wild type (WT) data. In order to answer a question whether experimental data contain enough information for the correct prediction we introduce two measures of predictive power. The first measure reveals the biologically substantiated low sensitivity of the model to parameters that are responsible for correct reconstruction of expression patterns in mutants, while the second one takes into account their correlation with the other parameters. It is demonstrated that the model solution, obtained by fitting to gene expression data in WT and Kr^- mutants simultaneously, and exhibiting the high predictive power, is characterized by much higher values of both measures than those fitted to WT data alone. This result leads us to the conclusion that information contained in WT data is insufficient to reliably estimate the large number of model parameters and provide predictions of mutants.

The temperature in the Arctic region has been increasing in the recent past accompanied by melting of its glaciers. We took a snapshot of the current microbial inhabitation of an Alaskan glacier (which can be considered as one of the simplest possible ecosystems) by using metagenomic sequencing of 16S rRNA recovered from ice/snow samples. Somewhat contrary to our expectations and earlier estimates, a rich and diverse microbial population of more than 2,500 species was revealed including several species of Archaea that has been identified for the first time in the glaciers of the Northern hemisphere. The most prominent bacterial groups found were Proteobacteria, Bacteroidetes, and Firmicutes. Firmicutes were not reported in large numbers in a previously studied Alpine glacier but were dominant in an Antarctic subglacial lake. Representatives of Cyanobacteria, Actinobacteria and Planctomycetes were among the most numerous, likely reflecting the dependence of the ecosystem on the energy obtained through photosynthesis and close links with the microbial community of the soil. Principal component analysis (PCA) of nucleotide word frequency revealed distinct sequence clusters for different taxonomic groups in the Alaskan glacier community and separate clusters for the glacial communities from other regions of the world. Comparative analysis of the community composition and bacterial diversity present in the Byron glacier in Alaska with other environments showed larger overlap with an Arctic soil than with a high Arctic lake, indicating patterns of community exchange and suggesting that these bacteria may play an important role in soil development during glacial retreat.

Reconstruction of phylogeny of a protein family from a sequence alignment can produce results of different quality. Our goal is to predict the quality of phylogeny reconstruction basing on features that can be extracted from the input alignment. We used Fitch–Margoliash (FM) method of phylogeny reconstruction and random forest as a predictor. For training and testing the predictor, alignments of orthologous series (OS) were used, for which the result of phylogeny reconstruction can be evaluated by comparison with trees of corresponding organisms. Our results show that the quality of phylogeny reconstruction can be predicted with more than 80% precision. Also, we tried to predict which phylogeny reconstruction method, FM or UPGMA, is better for a particular alignment. With the used set of features, among alignments for which the obtained predictor predicts a better performance of UPGMA, 56% really give a better result with UPGMA. Taking into account that in our testing set only for 34% alignments UPGMA performs better, this result shows a principal possibility to predict the better phylogeny reconstruction method basing on features of a sequence alignment.

Using the combination of molecular dynamics (MD) simulations and geometric clustering we analyzed the role of arginine at 209 position in the transition of protein kinase A Iα (PKA Iα) regulatory subunit A-domain from H- to B-conformation and stabilization of the latter. The mechanism underlying the role of the residue at position 209 in the realization of B-conformation includes: (1) possibility to bind the ligand tightly (if transition happens in the presence of cAMP), (2) capability to hold β2β3-loop in the correct conformation, (3) tendency of residue at 209 position to stabilize B-conformation in the absence and in presence of the ligand. In terms of the effect produced on transition of A-domain from H- to B-conformation in the presence of cAMP, mutational substitutions for R209 can be arranged in the following order: Glu(Gly)>Lys>Ile. In the absence of cAMP the order is different Lys>Gly>Glu>Ile. Thus, our results allow us to presume that the role of arginine at 209 position can be important though not crucial.

Seventy-eight promoter islands with an extraordinarily high density of potential promoters have been recently found in the genome of Escherichia coli. It has been shown that RNA polymerase binds internal promoters of these islands and produces short oligonucleotides, while the synthesis of normal mRNAs is suppressed. This quenching may be biologically relevant, as most islands are associated with foreign genes, which expression may deplete cellular resources. However, a molecular mechanism of silencing with the participation of these promoter-rich regions remains obscure. It has been demonstrated that all islands interact with histone-like protein H-NS — a specific sentinel of foreign genes. In this study, we demonstrated the inhibitory effect of H-NS using Δhns mutant of Escherichia coli and showed that deletion of dps, encoding another protein of bacterial nucleoid, tended to decrease rather than increase the amount of island-specific transcripts. This observation precluded consideration of promoter islands as sites for targeted heterochromatization only and a computer search for the binding sites of 53 transcription factors (TFs) revealed six proteins, which may specifically regulate their transcriptional output.

Chemogenomic experiments, where genetic and chemical perturbations are combined, provide data for discovering the relationships between genotype and phenotype. Traditionally, analysis of chemogenomic datasets has been done considering the sensitivity of the deletion strains to chemicals, and this has shed light on drug mechanism of action and detecting drug targets. Here, we computationally analyzed a large chemogenomic dataset, which combines more than 300 chemicals with virtually all gene deletion strains in the yeast S. cerevisiae. In addition to sensitivity relation between deletion strains and chemicals, we also considered the deletion strains that are resistant to chemicals. We found a small set of genes whose deletion makes the cell resistant to many chemicals. Curiously, these genes were enriched for functions related to RNA metabolism. Our approach allowed us to generate a network of drugs and genes that are connected with resistance or sensitivity relationships. As a quality assessment, we showed that the higher order motifs found in this network are consistent with biological expectations. Finally, we constructed a biologically relevant network projection pertaining to drug similarities, and analyzed this network projection in detail. We propose this drug similarity network as a useful tool for understanding drug mechanism of action.

Interactions between integral membrane proteins hemagglutinin (HA), neuraminidase (NA), M2 and membrane-associated matrix protein M1 of influenza A virus are thought to be crucial for assembly of functionally competent virions. We hypothesized that the amino acid residues located at the interface of two different proteins are under physical constraints and thus probably co-evolve. To predict co-evolving residue pairs, the EvFold (http://evfold.org) program searching the (nontransitive) Direct Information scores was applied for large samplings of amino acid sequences from Influenza Research Database (http://www.fludb.org/). Having focused on the HA, NA, and M2 cytoplasmic tails as well as C-terminal domain of M1 (being the less conserved among the protein domains) we captured six pairs of correlated positions. Among them, there were one, two, and three position pairs for HA–M2, HA–M1, and M2–M1 protein pairs, respectively. As expected, no co-varying positions were found for NA–HA, NA–M1, and NA–M2 pairs obviously due to high conservation of the NA cytoplasmic tail. The sum of frequencies calculated for two major amino acid patterns observed in pairs of correlated positions was up to 0.99 meaning their high to extreme evolutionary sustainability. Based on the predictions a hypothetical model of pair-wise protein interactions within the viral envelope was proposed.

Biological development depends on the coordinated expression of genes in time and space. Developmental genes have extensive cis-regulatory regions which control their expression. These regions are organized in a modular manner, with different modules controlling expression at different times and locations. Both how modularity evolved and what function it serves are open questions. We present a computational model for the cis-regulation of the hunchback (hb) gene in the fruit fly (Drosophila). We simulate evolution (using an evolutionary computation approach from computer science) to find the optimal cis-regulatory arrangements for fitting experimental hb expression patterns. We find that the cis-regulatory region tends to readily evolve modularity. These cis-regulatory modules (CRMs) do not tend to control single spatial domains, but show a multi-CRM/multi-domain correspondence. We find that the CRM-domain correspondence seen in Drosophila evolves with a high probability in our model, supporting the biological relevance of the approach. The partial redundancy resulting from multi-CRM control may confer some biological robustness against corruption of regulatory sequences. The technique developed on hb could readily be applied to other multi-CRM developmental genes.

Most of the associated single nucleotide polymorphisms (SNPs) for genome wide association studies (GWAS) explain very little proportion of phenotypic variance in outbred populations. One reason is; large number of markers raises the problem of multiple hypothesis testing correction using conservative statistical tests in single marker models. Admixture mapping could be used as alternative model to detect the genes associated with quantitative traits by less number of ancestry informative markers. Ancestral genotypes of founder populations were available for the F₂ mice dataset for growth related traits. The objectives of this study were (1) to detect genomic signals by admixture mapping for growth related traits by ancestry informative markers and ancestral genotypes (2) to detect genomic signals for growth related traits by Bayes C(π) model and compare results with those obtained by use of admixture mapping. Bayes C(π) model detected more SNPs that has high ancestry informative markers. But due to stringent significance tests and small SNPs effects admixture model did not detect the same SNPs in Bayes C(π). As was expected higher ancestral informative markers lead to higher Z values in admixture model with a little variation. Admixture model could incorporate and use ancestral genomic information.