Russell Greiner

HMDB 3.0—The Human Metabolome Database in 2013

The Human Metabolome Database (HMDB) (www.hmdb.ca) is a resource dedicated to providing scientists with the most current and comprehensive coverage of the human metabolome. Since its first release in 2007, the HMDB has been used to facilitate research for nearly 1000 published studies in metabolomics, clinical biochemistry and systems biology. The most recent release of HMDB (version 3.0) has been significantly expanded and enhanced over the 2009 release (version 2.0). In particular, the number of annotated metabolite entries has grown from 6500 to more than 40 000 (a 600% increase). This enormous expansion is a result of the inclusion of both ‘detected’ metabolites (those with measured concentrations or experimental confirmation of their existence) and ‘expected’ metabolites (those for which biochemical pathways are known or human intake/exposure is frequent but the compound has yet to be detected in the body). The latest release also has greatly increased the number of metabolites with biofluid or tissue concentration data, the number of compounds with reference spectra and the number of data fields per entry. In addition to this expansion in data quantity, new database visualization tools and new data content have been added or enhanced. These include better spectral viewing tools, more powerful chemical substructure searches, an improved chemical taxonomy and better, more interactive pathway maps. This article describes these enhancements to the HMDB, which was previously featured in the 2009 NAR Database Issue. (Note to referees, HMDB 3.0 will go live on 18 September 2012.).

HMDB: the Human Metabolome Database

The Human Metabolome Database (HMDB) is currently the most complete and comprehensive curated collection of human metabolite and human metabolism data in the world. It contains records for more than 2180 endogenous metabolites with information gathered from thousands of books, journal articles and electronic databases. In addition to its comprehensive literature-derived data, the HMDB also contains an extensive collection of experimental metabolite concentration data compiled from hundreds of mass spectra (MS) and Nuclear Magnetic resonance (NMR) metabolomic analyses performed on urine, blood and cerebrospinal fluid samples. This is further supplemented with thousands of NMR and MS spectra collected on purified, reference metabolites. Each metabolite entry in the HMDB contains an average of 90 separate data fields including a comprehensive compound description, names and synonyms, structural information, physico-chemical data, reference NMR and MS spectra, biofluid concentrations, disease associations, pathway information, enzyme data, gene sequence data, SNP and mutation data as well as extensive links to images, references and other public databases. Extensive searching, relational querying and data browsing tools are also provided. The HMDB is designed to address the broad needs of biochemists, clinical chemists, physicians, medical geneticists, nutritionists and members of the metabolomics community. The HMDB is available at:

The Human Serum Metabolome

Continuing improvements in analytical technology along with an increased interest in performing comprehensive, quantitative metabolic profiling, is leading to increased interest pressures within the metabolomics community to develop centralized metabolite reference resources for certain clinically important biofluids, such as cerebrospinal fluid, urine and blood. As part of an ongoing effort to systematically characterize the human metabolome through the Human Metabolome Project, we have undertaken the task of characterizing the human serum metabolome. In doing so, we have combined targeted and non-targeted NMR, GC-MS and LC-MS methods with computer-aided literature mining to identify and quantify a comprehensive, if not absolutely complete, set of metabolites commonly detected and quantified (with todays technology) in the human serum metabolome. Our use of multiple metabolomics platforms and technologies allowed us to substantially enhance the level of metabolome coverage while critically assessing the relative strengths and weaknesses of these platforms or technologies. Tables containing the complete set of 4229 confirmed and highly probable human serum compounds, their concentrations, related literature references and links to their known disease associations are freely available at http://www.serummetabolome.ca.

Learning Bayesian networks from data: an information-theory based approach

This paper provides algorithms that use an information-theoretic analysis to learn Bayesian network structures from data. Based on our three-phase learning framework, we develop efficient algorithms that can effectively learn Bayesian networks, requiring only polynomial numbers of conditional independence (CI) tests in typical cases. We provide precise conditions that specify when these algorithms are guaranteed to be correct as well as empirical evidence (from real world applications and simulation tests) that demonstrates that these systems work efficiently and reliably in practice.

BASys: a web server for automated bacterial genome annotation

BASys (Bacterial Annotation System) is a web server that supports automated, in-depth annotation of bacterial genomic (chromosomal and plasmid) sequences. It accepts raw DNA sequence data and an optional list of gene identification information and provides extensive textual annotation and hyperlinked image output. BASys uses >30 programs to determine ∼60 annotation subfields for each gene, including gene/protein name, GO function, COG function, possible paralogues and orthologues, molecular weight, isoelectric point, operon structure, subcellular localization, signal peptides, transmembrane regions, secondary structure, 3D structure, reactions and pathways. The depth and detail of a BASys annotation matches or exceeds that found in a standard SwissProt entry. BASys also generates colorful, clickable and fully zoomable maps of each query chromosome to permit rapid navigation and detailed visual analysis of all resulting gene annotations. The textual annotations and images that are provided by BASys can be generated in ∼24 h for an average bacterial chromosome (5 Mb). BASys annotations may be viewed and downloaded anonymously or through a password protected access system. The BASys server and databases can also be downloaded and run locally. BASys is accessible at .

A correction to the algorithm in Reiter's theory of diagnosis

Reiter [3] has developed a general theory of diagnosis based on first principles. His algorithm computes all diagnoses which explain the differences between the predicted and observed behavior of a given system. Unfortunately, Reiters description of the algorithm is incorrect in that some diagnoses can be missed under certain conditions. This note presents a revised algorithm and a proof of its correctness.

Learning Bayesian Belief Network Classifiers: Algorithms and System

This paper investigates the methods for learning predictive classifiers based on Bayesian belief networks (BN) - primarily unrestricted Bayesian networks and Bayesian multi-nets. We present our algorithms for learning these classifiers, and discuss how these methods address the overfitting problem and provide a natural method for feature subset selection. Using a set of standard classification problems, we empirically evaluate the performance of various BN-based classifiers. The results show that the proposed BN and Bayes multinet classifiers are competitive with (or superior to) the best known classifiers, based on both BN and other formalisms; and that the computational time for learning and using these classifiers is relatively small. These results argue that BN-based classifiers deserve more attention in the data mining community.

The human cerebrospinal fluid metabolome.

With continuing improvements in analytical technology and an increased interest in comprehensive metabolic profiling of biofluids and tissues, there is a growing need to develop comprehensive reference resources for certain clinically important biofluids, such as blood, urine and cerebrospinal fluid (CSF). As part of our effort to systematically characterize the human metabolome we have chosen to characterize CSF as the first biofluid to be intensively scrutinized. In doing so, we combined comprehensive NMR, gas chromatography-mass spectrometry (GC-MS) and liquid chromatography (LC) Fourier transform-mass spectrometry (FTMS) methods with computer-aided literature mining to identify and quantify essentially all of the metabolites that can be commonly detected (with todays technology) in the human CSF metabolome. Tables containing the compounds, concentrations, spectra, protocols and links to disease associations that we have found for the human CSF metabolome are freely available at http://www.csfmetabolome.ca.

Structural extension to logistic regression: discriminative parameter learning of belief net classifiers

Bayesian belief nets (BNs) are often used for classification tasks—typically to return the most likely class label for each specified instance. Many BN-learners, however, attempt to find the BN that maximizes a different objective function—viz., likelihood, rather than classification accuracy—typically by first learning an appropriate graphical structure, then finding the parameters for that structure that maximize the likelihood of the data. As these parameters may not maximize the classification accuracy, “discriminative parameter learners” follow the alternative approach of seeking the parameters that maximize conditional likelihood (CL), over the distribution of instances the BN will have to classify. This paper first formally specifies this task, shows how it extends standard logistic regression, and analyzes its inherent sample and computational complexity. We then present a general algorithm for this task, ELR, that applies to arbitrary BN structures and that works effectively even when given incomplete training data. Unfortunately, ELR is not guaranteed to find the parameters that optimize conditional likelihood; moreover, even the optimal-CL parameters need not have minimal classification error. This paper therefore presents empirical evidence that ELR produces effective classifiers, often superior to the ones produced by the standard “generative” algorithms, especially in common situations where the given BN-structure is incorrect.

The COMBREX Project: Design, Methodology, and Initial Results

Experimental data exists for only a vanishingly small fraction of sequenced microbial genes. This community page discusses the progress made by the COMBREX project to address this important issue using both computational and experimental resources.