Mario Latendresse

EcoCyc: fusing model organism databases with systems biology

EcoCyc (http://EcoCyc.org) is a model organism database built on the genome sequence of Escherichia coli K-12 MG1655. Expert manual curation of the functions of individual E. coli gene products in EcoCyc has been based on information found in the experimental literature for E. coli K-12-derived strains. Updates to EcoCyc content continue to improve the comprehensive picture of E. coli biology. The utility of EcoCyc is enhanced by new tools available on the EcoCyc web site, and the development of EcoCyc as a teaching tool is increasing the impact of the knowledge collected in EcoCyc.

Pathway Tools version 13.0: integrated software for pathway/genome informatics and systems biology

Pathway Tools is a production-quality software environment for creating a type of model-organism database called a Pathway/Genome Database (PGDB). A PGDB such as EcoCyc integrates the evolving understanding of the genes, proteins, metabolic network and regulatory network of an organism. This article provides an overview of Pathway Tools capabilities. The software performs multiple computational inferences including prediction of metabolic pathways, prediction of metabolic pathway hole fillers and prediction of operons. It enables interactive editing of PGDBs by DB curators. It supports web publishing of PGDBs, and provides a large number of query and visualization tools. The software also supports comparative analyses of PGDBs, and provides several systems biology analyses of PGDBs including reachability analysis of metabolic networks, and interactive tracing of metabolites through a metabolic network. More than 800 PGDBs have been created using Pathway Tools by scientists around the world, many of which are curated DBs for important model organisms. Those PGDBs can be exchanged using a peer-to-peer DB sharing system called the PGDB Registry.

EcoCyc: a comprehensive database of Escherichia coli biology

EcoCyc (http://EcoCyc.org) is a comprehensive model organism database for Escherichia coli K-12 MG1655. From the scientific literature, EcoCyc captures the functions of individual E. coli gene products; their regulation at the transcriptional, post-transcriptional and protein level; and their organization into operons, complexes and pathways. EcoCyc users can search and browse the information in multiple ways. Recent improvements to the EcoCyc Web interface include combined gene/protein pages and a Regulation Summary Diagram displaying a graphical overview of all known regulatory inputs to gene expression and protein activity. The graphical representation of signal transduction pathways has been updated, and the cellular and regulatory overviews were enhanced with new functionality. A specialized undergraduate teaching resource using EcoCyc is being developed.

Multidimensional annotation of the Escherichia coli K-12 genome

The annotation of the Escherichia coli K-12 genome in the EcoCyc database is one of the most accurate, complete and multidimensional genome annotations. Of the 4460 E. coli genes, EcoCyc assigns biochemical functions to 76%, and 66% of all genes had their functions determined experimentally. EcoCyc assigns E. coli genes to Gene Ontology and to MultiFun. Seventy-five percent of gene products contain reviews authored by the EcoCyc project that summarize the experimental literature about the gene product. EcoCyc information was derived from 15 000 publications. The database contains extensive descriptions of E. coli cellular networks, describing its metabolic, transport and transcriptional regulatory processes. A comparison to genome annotations for other model organisms shows that the E. coli genome contains the most experimentally determined gene functions in both relative and absolute terms: 2941 (66%) for E. coli, 2319 (37%) for Saccharomyces cerevisiae, 1816 (5%) for Arabidopsis thaliana, 1456 (4%) for Mus musculus and 614 (4%) for Drosophila melanogaster. Database queries to EcoCyc survey the global properties of E. coli cellular networks and illuminate the extent of information gaps for E. coli, such as dead-end metabolites. EcoCyc provides a genome browser with novel properties, and a novel interactive display of transcriptional regulatory networks.

Creation of a Genome-Wide Metabolic Pathway Database for Populus trichocarpa Using a New Approach for Reconstruction and Curation of Metabolic Pathways for Plants

Metabolic networks reconstructed from sequenced genomes or transcriptomes can help visualize and analyze large-scale experimental data, predict metabolic phenotypes, discover enzymes, engineer metabolic pathways, and study metabolic pathway evolution. We developed a general approach for reconstructing metabolic pathway complements of plant genomes. Two new reference databases were created and added to the core of the infrastructure: a comprehensive, all-plant reference pathway database, PlantCyc, and a reference enzyme sequence database, RESD, for annotating metabolic functions of protein sequences. PlantCyc (version 3.0) includes 714 metabolic pathways and 2,619 reactions from over 300 species. RESD (version 1.0) contains 14,187 literature-supported enzyme sequences from across all kingdoms. We used RESD, PlantCyc, and MetaCyc (an all-species reference metabolic pathway database), in conjunction with the pathway prediction software Pathway Tools, to reconstruct a metabolic pathway database, PoplarCyc, from the recently sequenced genome of Populus trichocarpa. PoplarCyc (version 1.0) contains 321 pathways with 1,807 assigned enzymes. Comparing PoplarCyc (version 1.0) with AraCyc (version 6.0, Arabidopsis [Arabidopsis thaliana]) showed comparable numbers of pathways distributed across all domains of metabolism in both databases, except for a higher number of AraCyc pathways in secondary metabolism and a 1.5-fold increase in carbohydrate metabolic enzymes in PoplarCyc. Here, we introduce these new resources and demonstrate the feasibility of using them to identify candidate enzymes for specific pathways and to analyze metabolite profiling data through concrete examples. These resources can be searched by text or BLAST, browsed, and downloaded from our project Web site (http://plantcyc.org).

Behavioral contracts and behavioral subtyping

Component-based software manufacturing has the potential to bring division-of-labor benefits to the world of software engineering. In order to make a market of software components viable, however, producers and consumers must agree on enforceable software contracts. In this paper, we show how to enforce contracts if components are manufactured from class and interface hierarchies. In particular, we focus on one style of contract: pre- and post-conditions. Programmers annotate class and interface methods with pre- and post-conditions and the run-time system checks these conditions during evaluation. These contracts guarantee that methods are called properly and provide appropriate results. In procedural languages, the use of pre- and post-condition contracts is well-established and studies have demonstrated its value. In object-oriented languages, however, assigning blame for pre- and post-condition failures poses subtle and complex problems. Specifically, assigning blame for malformed class and interface hierarchies is so difficult that none of the existing contract monitoring tools correctly assign blame for these failures. In this paper, we show how to overcome these problems in the context of Java. Our work is based on the notion of behavioral subtyping.

Construction and completion of flux balance models from pathway databases

Motivation: Flux balance analysis (FBA) is a well-known technique for genome-scale modeling of metabolic flux. Typically, an FBA formulation requires the accurate specification of four sets: biochemical reactions, biomass metabolites, nutrients and secreted metabolites. The development of FBA models can be time consuming and tedious because of the difficulty in assembling completely accurate descriptions of these sets, and in identifying errors in the composition of these sets. For example, the presence of a single non-producible metabolite in the biomass will make the entire model infeasible. Other difficulties in FBA modeling are that model distributions, and predicted fluxes, can be cryptic and difficult to understand. Results: We present a multiple gap-filling method to accelerate the development of FBA models using a new tool, called MetaFlux, based on mixed integer linear programming (MILP). The method suggests corrections to the sets of reactions, biomass metabolites, nutrients and secretions. The method generates FBA models directly from Pathway/Genome Databases. Thus, FBA models developed in this framework are easily queried and visualized using the Pathway Tools software. Predicted fluxes are more easily comprehended by visualizing them on diagrams of individual metabolic pathways or of metabolic maps. MetaFlux can also remove redundant high-flux loops, solve FBA models once they are generated and model the effects of gene knockouts. MetaFlux has been validated through construction of FBA models for Escherichia coli and Homo sapiens. Availability: Pathway Tools with MetaFlux is freely available to academic users, and for a fee to commercial users. Download from: biocyc.org/download.shtml. Contact: mario.latendresse@sri.com Supplementary information: Supplementary data are available at Bioinformatics online.

The Pathway Tools Pathway Prediction Algorithm

The PathoLogic component of the Pathway Tools software performs prediction of metabolic pathways in sequenced and annotated genomes. This article provides a detailed presentation of the PathoLogic algorithm. The algorithm consists of two phases. The reactome inference phase infers the reactions catalyzed by the organism from the set of enzymes present in the annotated genome. The pathway inference phase infers the metabolic pathways present in the organism from the reactions catalyzed by the organism. Both phases draw on the MetaCyc database of metabolic reactions and pathways. MetaCyc contains two data fields to support pathway inference: the expected taxonomic range of each pathway, and a list of key reactions for pathways. These fields have significantly increased the predictive accuracy of PathoLogic.

The EcoCyc database: reflecting new knowledge about Escherichia coli K-12

EcoCyc (EcoCyc.org) is a freely accessible, comprehensive database that collects and summarizes experimental data for Escherichia coli K-12, the best-studied bacterial model organism. New experimental discoveries about gene products, their function and regulation, new metabolic pathways, enzymes and cofactors are regularly added to EcoCyc. New SmartTable tools allow users to browse collections of related EcoCyc content. SmartTables can also serve as repositories for user- or curator-generated lists. EcoCyc now supports running and modifying E. coli metabolic models directly on the EcoCyc website.

Accurate Atom-Mapping Computation for Biochemical Reactions

The complete atom mapping of a chemical reaction is a bijection of the reactant atoms to the product atoms that specifies the terminus of each reactant atom. Atom mapping of biochemical reactions is useful for many applications of systems biology, in particular for metabolic engineering where synthesizing new biochemical pathways has to take into account for the number of carbon atoms from a source compound that are conserved in the synthesis of a target compound. Rapid, accurate computation of the atom mapping(s) of a biochemical reaction remains elusive despite significant work on this topic. In particular, past researchers did not validate the accuracy of mapping algorithms. We introduce a new method for computing atom mappings called the minimum weighted edit-distance (MWED) metric. The metric is based on bond propensity to react and computes biochemically valid atom mappings for a large percentage of biochemical reactions. MWED models can be formulated efficiently as Mixed-Integer Linear Programs (MILPs). We have demonstrated this approach on 7501 reactions of the MetaCyc database for which 87% of the models could be solved in less than 10 s. For 2.1% of the reactions, we found multiple optimal atom mappings. We show that the error rate is 0.9% (22 reactions) by comparing these atom mappings to 2446 atom mappings of the manually curated Kyoto Encyclopedia of Genes and Genomes (KEGG) RPAIR database. To our knowledge, our computational atom-mapping approach is the most accurate and among the fastest published to date. The atom-mapping data will be available in the MetaCyc database later in 2012; the atom-mapping software will be available within the Pathway Tools software later in 2012.