Genome-scale metabolic models bridge the gap between genome-derived biochemical information and metabolic phenotypes in a principled manner, providing a solid interpretative framework for experimental data related to metabolic states, and enabling simple in silico experiments with whole-cell metabolism. Models have been reconstructed for almost 20 bacterial species, so far mainly through expert curation efforts integrating information from the literature with genome annotation. A wide variety of computational methods exploiting metabolic models have been developed and applied to bacteria, yielding valuable insights into bacterial metabolism and evolution, and providing a sound basis for computer-assisted design in metabolic engineering. Recent advances in computational systems biology and high-throughput experimental technologies pave the way for the systematic reconstruction of metabolic models from genomes of new species, and a corresponding expansion of the scope of their applications. In this review, we provide an introduction to the key ideas of metabolic modeling, survey the methods, and resources that enable model reconstruction and refinement, and chart applications to the investigation of global properties of metabolic systems, the interpretation of experimental results, and the re-engineering of their biochemical capabilities.

Crossover (CO) is a key process for the accurate segregation of homologous chromosomes during the first meiotic division. In most eukaryotes, meiotic recombination is not homogeneous along the chromosomes, suggesting a tight control of the location of recombination events. We genotyped 71 single nucleotide polymorphisms (SNPs) covering the entire chromosome 4 of Arabidopsis thaliana on 702 F2 plants, representing 1404 meioses and allowing the detection of 1171 COs, to study CO localization in a higher plant. The genetic recombination rates varied along the chromosome from 0 cM/Mb near the centromere to 20 cM/Mb on the short arm next to the NOR region, with a chromosome average of 4.6 cM/Mb. Principal component analysis showed that CO rates negatively correlate with the G+C content (P =3x10(-4)), in contrast to that reported in other eukaryotes. COs also significantly correlate with the density of single repeats and the CpG ratio, but not with genes, pseudogenes, transposable elements, or dispersed repeats. Chromosome 4 has, on average, 1.6 COs per meiosis, and these COs are subjected to interference. A detailed analysis of several regions having high CO rates revealed "hot spots" of meiotic recombination contained in small fragments of a few kilobases. Both the intensity and the density of these hot spots explain the variation of CO rates along the chromosome

Escherichia coli exhibits a wide range of lifestyles encompassing commensalism and various pathogenic behaviors which its highly dynamic genome contributes to develop. How environmental and host factors shape the genetic structure of E. coli strains remains, however, largely unknown. Following a previous study of E. coli genomic diversity, we investigated its diversity at the metabolic level by building and analyzing the genomescale metabolic networks of 29 E. coli strains (8 commensal and 21 pathogenic strains, including 6 Shigella strains). Using a tailor-made reconstruction strategy, we significantly improved the completeness and accuracy of the metabolic networks over default automatic reconstruction processes. Among the 1,545 reactions forming E. coli panmetabolism, 885 reactions were common to all strains. This high proportion of core reactions (57%) was found to be in sharp contrast to the low proportion (13%) of core genes in the E. coli pangenome, suggesting less diversity of metabolic functions compared to that of all gene functions. Core reactions were significantly overrepresented among biosynthetic reactions compared to the more variable degradation processes. Differences between metabolic networks were found to follow E. coli phylogeny rather than pathogenic phenotypes, except for Shigella networks, which were significantly more distant from the others. This suggests that most metabolic changes in non-Shigella strains were not driven by their pathogenic phenotypes. Using a supervised method, we were yet able to identify small sets of reactions related to pathogenicity or commensalism. The quality of our reconstructed networks also makes them reliable bases for building metabolic models.Escherichia coli is a versatile species encompassing commensal organisms, as well as intraintestinal E. coli (InPEc) and extraintestinal E. coli (ExPEc) pathogens (27, 49). This variety of lifestyles has been seen as a consequence of the huge E. coli genome plasticity (51). However, linking genomic elements to phenotypic behaviors is not trivial because several layers of biological processes separate genes from their phenotypic effects, and in extreme cases, the evolutionary path can lead either to the functional convergence of distinct sets of genes or to the functional divergence of an initially common set of genes. Consequently, in order to establish links between genomes and phenotypes, one needs an integrative layer. A recent study on a set of 20 E. coli strains (51) has shown that a large fraction of the shared genomic elements with known function is related to metabolism. Because it is now feasible to reconstruct metabolic networks at the genome scale (7,13,16,26), these metabolic networks can, in principle, be used as functional bridges between genomic diversity and phenotypic differences. Currently, such reconstructions are performed automatically from the annotation of input genomes, using algorithms that match these annotations with the contents of reference metabolic databases (13,16).In this work, we stu...

Bacteriophage genomes harbor the broadest chemical diversity of nucleobases across all life forms. Certain DNA viruses that infect hosts as diverse as cyanobacteria, proteobacteria, and actinobacteria exhibit wholesale substitution of aminoadenine for adenine, thereby forming three hydrogen bonds with thymine and violating Watson-Crick pairing rules. Aminoadenine-encoded DNA polymerases, homologous to the Klenow fragment of bacterial DNA polymerase I that includes 3′-exonuclease but lacks 5′-exonuclease, were found to preferentially select for aminoadenine instead of adenine in deoxynucleoside triphosphate incorporation templated by thymine. Polymerase genes occur in synteny with genes for a biosynthesis enzyme that produces aminoadenine deoxynucleotides in a wide array of Siphoviridae bacteriophages. Congruent phylogenetic clustering of the polymerases and biosynthesis enzymes suggests that aminoadenine has propagated in DNA alongside adenine since archaic stages of evolution.

Motivation: The construction of statistics for summarizing posterior samples returned by a Bayesian phylogenetic study has so far been hindered by the poor geometric insights available into the space of phylogenetic trees, and ad hoc methods such as the derivation of a consensus tree makeup for the ill-definition of the usual concepts of posterior mean, while bootstrap methods mitigate the absence of a sound concept of variance. Yielding satisfactory results with sufficiently concentrated posterior distributions, such methods fall short of providing a faithful summary of posterior distributions if the data do not offer compelling evidence for a single topology.Results: Building upon previous work of Billera et al., summary statistics such as sample mean, median and variance are defined as the geometric median, Fréchet mean and variance, respectively. Their computation is enabled by recently published works, and embeds an algorithm for computing shortest paths in the space of trees. Studying the phylogeny of a set of plants, where several tree topologies occur in the posterior sample, the posterior mean balances correctly the contributions from the different topologies, where a consensus tree would be biased. Comparisons of the posterior mean, median and consensus trees with the ground truth using simulated data also reveals the benefits of a sound averaging method when reconstructing phylogenetic trees.Availability and implementation: We provide two independent implementations of the algorithm for computing Fréchet means, geometric medians and variances in the space of phylogenetic trees. TFBayes: https://github.com/pbenner/tfbayes, TrAP: https://github.com/bacak/TrAP.Contact: philipp.benner@mis.mpg.de

The Mixture Transition Distribution (MTD) model was introduced by Raftery to face the need for parsimony in the modeling of high-order Markov chains in discrete time. The particularity of this model comes from the fact that the effect of each lag upon the present is considered separately and additively, so that the number of parameters required is drastically reduced. However, the efficiency for the MTD parameter estimations proposed up to date still remains problematic on account of the large number of constraints on the parameters. In this paper, an iterative procedure, commonly known as Expectation-Maximization (EM) algorithm, is developed cooperating with the principle of Maximum Likelihood Estimation (MLE) to estimate the MTD parameters. Some applications of modeling MTD show the proposed EM algorithm is easier to be used than the algorithm developed by Berchtold. Moreover, the EM Estimations of parameters for high-order MTD models led on DNA sequences outperform the corresponding fully parametrized Markov chain in terms of Bayesian Information Criterion.A software implementation of our algorithm is available in the library seq++ at http://stat.genopole.cnrs.fr/seqpp.

Abstract. We introduce inhomogeneous parsimonious Markov models for modeling statistical patterns in discrete sequences. These models are based on parsimonious context trees, which are a generalization of context trees, and thus generalize variable order Markov models. We follow a Bayesian approach, consisting of structure and parameter learning. Structure learning is a challenging problem due to an overexponential number of possible tree structures, so we describe an exact and efficient dynamic programming algorithm for finding the optimal tree structures.We apply model and learning algorithm to the problem of modeling binding sites of the human transcription factor C/EBP, and find an increased prediction performance compared to fixed order and variable order Markov models. We investigate the reason for this improvement and find several instances of context-specific dependences that can be captured by parsimonious context trees but not by traditional context trees.

The seq++ package offers a reference set of programs and an extensible library to biologists and developers working on sequence statistics. Its generality arises from the ability to handle sequences described with any alphabet (nucleotides, amino acids, codons and others). seq++ enables sequence modelling with various types of Markov models, including variable length Markov models and the newly developed parsimonious Markov models, all of them potentially phased. Simulation modules are supplied for Monte Carlo methods. Hence, this toolbox allows the study of any biological process which can be described by a series of states taken from a finite set.