MeltDB

Category Metabolomics/Metabonomics>Knowledge Bases/Databases/Tools and Metabolomics/Metabonomics>Metabolic Profiling/Analysis Systems/Tools

Abstract MeltDB is a web based framework for the storage, analysis and visualization of metabolomic datasets and their integration with experimental data from transcriptomics, genomics, and proteomics.

MeltDB supports open file formats (netCDF, mzXML, mzDATA) and facilitates the integration and evaluation of existing preprocessing methods. The system provides researchers with the means to consistently describe and store their experimental datasets.

Comprehensive analysis and visualization features of metabolomics datasets are offered to the community through a web-based user interface.

The system covers the process from raw data to the visualization of results in a knowledge-based background and is integrated into the context of existing software platforms of genomics and transcriptomics at Bielefeld University.

MeltDB is designed to provide analysis methods for raw Gas Chromatography (GC) - and Liquid Chromatography-Mass Spectrometry LC-MS datasets and offers methods to combine the respective results.

A flexible tool pipeline (workflow) implemented in MeltDB allows both the import of preprocessed data as well as the integration of existing open source analysis packages such as XCMS (XCMS - Processes Mass Spectrometry data for ‘Metabolite Profiling’ using Nonlinear Peak Alignment, Matching, and Identification), MassSpecWavelet or metaB.

For the identification of metabolites based on mass spectra the freely available Golm Metabolome Database (GMD) database is queried. Additionally, user defined libraries in the NIST format can be imported.

To facilitate the sound statistical analysis of preprocessed metabolite quantities, MeltDB supports normalization of metabolite quantities by internal standards (e.g. Ribitol) and also dry weight or cell volumina.

The integration of the R statistics software allows you to apply standard methods [T-Tests, Analysis of Variance (AOV), and Hierarchical Clustering], and generate explorative visualizations (PCA, ICA, Heatmaps Clustering results) on the normalized datasets.

The integration of genomic and transcriptomic datasets originating from GenDB (GenDB is an annotation system for ‘prokaryotic genomes’) or EMMA 2 (A MAGE-compliant system for the collaborative analysis and integration of microarray data) is achieved via SOAP based web services.

Thus, interactive visualizations of metabolite concentrations together with transcript measurements mapped on e.g. KEGG pathways can be easily generated.

MeltDB Data model --

In order to be able to support the different objectives of metabolomic experiments, four (4) specialized subclasses have been defined in accordance with the suggestions given by the Metabolomics Standards Initiative (MSI) workgroup:

1) Targeted analysis - detection and precise quantification of a single or a small set of target compounds within a metabolome sample.

2) Metabolite profiling - detection and approximate quantification of a large set of target metabolites within a metabolite sample.

3) Metabolomics - detection, approximate quantification and tentative identification of as many of the compounds within a metabolome sample as possible.

4) Metabolic fingerprinting - generation of a signature for a metabolome sample without regard for the individual compounds that it contains.

MeltDB Experiment description --

The MeltDB data model is able to describe the experimental design representing, e.g. growth conditions or extraction and sample preparation methods. Each chromatogram in MeltDB can be attributed with a list of these experimental factors which thereby also annotate the experimental conditions.

The structure of the best practice recommendations of the MSI working group is integrated and can be dynamically extended. Once experimental factors have been defined in the MeltDB database, they can be reused for the annotation of multiple chromatograms.

MeltDB Project management --

MeltDB is a web-based application and runs on a dedicated server that provides access to various projects. Fine grained project and user management is included in order to make the system applicable to datasets that are analyzed across more than one institute with various research groups working in parallel.

MeltDB Analysis Pipeline --

A pipeline (workflow) concept has been defined that allows the integration of existing tools such as XCMS and importers for results generated by proprietary software from the vendors of GC/LC-MS machinery.

For mass spectral database searches the algorithm proposed by Stein and Scott (1994) has been re-implemented and extended. Queries to the freely available GMD or user defined spectral databases in NIST or AMDIS format are supported.

Additionally, peak detection and quantification routines have been implemented. Retention Index calculation together with a Dynamic Time Warping approach for chromatogram alignment is provided and contributes to the improved identification of metabolites.

These methods form the basis of the preprocessing pipeline of MeltDB that can automatically generate normalized metabolite concentration matrices from raw chromatographic datasets.

MeltDB Tool concept --

MeltDB features a flexible tool concept that facilitates the integration of existing open source software packages from computational metabolomics. It may become necessary to adopt the parameterization of these tools to the datasets under study.

Thus, the generic Tool concept allows you to store multiple parameterizations of single methods and applications in the MeltDB database. The experimenter can select a suitable tool instance for the preprocessing of their chromatograms or experiments and submit the actual computation to a compute cluster.

MeltDB Importer --

For preprocessing methods that cannot be employed in an open source environment, integration can be achieved through specialized importing functionality, using the previously described tool concept. Importers have been implemented for the following ‘file formats’:

The importers transform information on the chromatographic peaks predicted by external software tools into the MeltDB data model. Information associated with peaks such as chemical identity, peak area and intensity are represented using observations and annotations.

MeltDB Visualization methods --

The availability of standardized interfaces to the different raw file formats provided by the MeltDB Application Programming Interface (API) allows you to realize generic visualizations, such as Total Ion Chromatogram (TIC) views and chromatogram alignments for whole experiments.

All visualizations are enriched with information on detected peaks in the MeltDB web interface. Interactive access to the underlying information stored in associated observations and annotations is also provided via the web interface.

MeltDB Statistical analysis --

Freely available R packages from BioConductor offer a multitude of analysis functions and visualizations developed for functional genomics datasets.

MeltDB database objects are dynamically converted to data objects in R in a standardized manner, which results in an integrated and efficient way to statistically analyze metabolomic experiments stored in MeltDB.

All statistical methods can be executed using the web interface and additional criteria, such as the treatment of missing values, the scaling of the values and the exclusion of certain metabolites or whole chromatograms can be controlled by the user.

The following list shows the statistical and explorative analysis methods currently available in MeltDB:

Additional analysis and visualization functionality provided through BioConductor packages or R can be easily added to MeltDB. Functionality can either be realized using the MeltDB Tool concept or data visualizations generated by R can be embedded in the MeltDB web interface.

MeltDB Data integration --

Data integration in MeltDB is achieved on various levels. The KEGG compound database is regularly imported through direct access to the KEGG FTP server. Relevant terms and relations of the compounds are directly represented in the MeltDB database model.

Since the manufacturer’s use the KEGG compound database as a controlled vocabulary for compounds and thereby obtain a connection of metabolites to pathways, enzymes and genes, it is straightforward to link metabolic experiments with existing genome projects stored in GenDB.

System Requirements

Web-based.

Manufacturer

Manufacturer Web Site MeltDB

Price Contact manufacturer.

G6G Abstract Number 20662

G6G Manufacturer Number 104306