CPSA Digest 2002

Proceedings -Wednesday, October 9, 2002

WeOE2

Emerging Approaches to Data Integration and Bioinformatics in Toxicology and Disease Studies

Background
While understanding the human genome has yielded significant insight into factors related to disease, it is only one piece of a complex system. This presentation goes beyond genomics and looks for correlations, pathways, and connections among thousands of measurable molecular components—including genes, proteins, and metabolites—in clinical samples. A Systems Biology approach is used to reveal unique disease markers and drug targets.

Systems Biology is the study of biology as an integrated system of genetic, protein, metabolite, cellular, and pathway events that are in flux and interdependent. While genes are informative clues to disease, they are not the active agents and insight into disease pathology is gained through knowledge of protein and metabolite levels. Beyond Genomics uses its proprietary technologies to perform parallel, comparative analyses of protein, metabolite and mRNA levels in complex biological samples, such as peripheral fluids and tissues.

The application to drug discovery for this technology includes utilizing clinical samples from diseased and healthy (normal) patients to uncover BioSystem Markers™ and BioSelective Targets™ that are indicators of disease and potential targets for therapeutic intervention. These markers and targets enable the linkage of gene response, protein activity and metabolite dynamics information to give rise to new pathway and system knowledge.

Premise
An integrated analytical approach lends itself to uniform experimental designs where sample collection methodologies can be standardized and sample handling and tracking are consistent. Often the same samples may be used for more than one type of analysis, i.e., the same tissue sample can be surveyed for metabolite, protein and mRNA profiles, and this enables both consistency in data normalization across platforms as well as complex correlation analyses that utilize profiling data pertinent to each biomolecule class. Analytical platforms are designed to be versatile, sensitive, and robust. Also important are dynamic range, coverage, and the ability to identify analytes of interest.

Metabolite profiling is achieved using both global NMR and LC/MS platforms that were developed to survey analytes that might be found at higher abundance in biological samples and more targeted LC/MS platforms that were designed to focus on particular classes of molecules that are lower in abundance. Beyond Genomics¹ approach includes both protein- and peptide-level separation strategies that allow for global and targeted coverage of the proteome, coupled either off-line or on-line to mass spectrometry for quantitative measurement. Also available are a number of platforms for quantitative profiling of proteins

Given the complexity of the datasets, where a single LC/MS chromatogram may contain peaks from thousands of analytes, it is essential to have informatics tools to extract relevant information from raw data. Beyond Genomics uses the IMPRESSTM algorithm for this data pre-processing step. The algorithm performs both a background noise removal step as well as a peak detection function that integrates all chromatographic peaks across all mass to charge ratios (m/z). Output includes peak list where each signal has a unique identifier that tracks chromatographic retention time and peak m/z, a measured intensity, and an assessment of the peak quality.

Following extraction of peak information, the next challenge is to analyze the data to determine if there are signals that differentiate groups of samples, e.g., a diseased population from a healthy population. The BioSystematics™ approach incorporates computational strategies that include: normalization of the datasets; exploratory principal component analyses (PCA) to identify differences between groups of samples; parametric and non-parametric tests of significance for a focused, signal-by-signal comparative analysis; and linear and non-linear correlation analyses to identify signal variances and changes that might be related biologically.

Analysis
Results from application of this methodology are shown below, where comparative metabolite, protein, mRNA expression profiling were performed in a study of a transgenic mouse model of atherosclerosis. The transgenic mice were sampled prior to the development of clinical symptoms of disease and compared against age-matched wild type controls in order to identify early biomarkers and to gain insight into pathogenesis. The figure shows a PCA analysis of LC/MS profiles from tryptic digests of a plasma protein fraction. The transgenic mice (TG) form a distinct cluster apart from the wild type controls (WT) and the difference factor spectrum indicates the differentially abundant tryptic peptides that give rise to the separation. These peptides are then targeted for sequencing via tandem mass spectrometry (MS/MS).

Value of the Technology
Systems biology requires differential comparative analyses that span a number of molecular classes and therefore a variety of robust analytical methods are necessary. Equally important are software tools for data warehousing and pre-processing, as well as bioinformatic approaches for data analysis and integration. When applied to disease studies, systems biology enables the discovery of biomarkers as well as potential avenues for therapeutic intervention.

References and/or Links

Beyond Genomics website at www.beyondgenomics.com

Return to Proceedings »