Research

Bioinformatics and Cancer Diagnosis

Over the last three decades, INCOGEN and its collaborators have played leading roles in efforts to develop the bioinformatics and technology infrastructure and provide scientists with analytic solution to enable effective research. Seeded by INCOGEN's first research grant, an award from the National Institute of Standards and Technology - Advanced Technology Program in 2001, the efforts culminated in numerous tools, such as the Visual Integrated Bioinformatics Environment (VIBE™) and GenePort®, that have assisted thousands of scientists worldwide in their research by providing effective integration of diverse and semantically heterogeneous data.

The last three decades have also seen an increased emphasis on system-level, integrated science as clinical researchers have recognized that the characterization of single genes and proteins has provided only limited insight and benefits toward early diagnoses, improved subtyping and prognoses, and treatment of complex, multifactorial diseases. This integrative approach is critical for our ability to elucidate the network of structural, regulatory, and dynamic interactions, thereby providing a comprehensive understanding of the physiology and pathophysiology that ultimately leads to effective intervention strategies.

Gene expression heatmap for ccRCC early vs late stage

Differentially expressed genes in clear cell renal cell carcinoma (ccRCC), early stage vs. late stage. Red - upregulated, green - downregulated.

The wide array of analysis tools and platforms available for analysis of clinical and life science data provide an unprecedented potential for discovery and benefits to health care. In particular, the application of multi-analyte molecular signatures derived from technologies such as protein mass spectrometry and gene expression microarrays has extraordinary potential for diagnostic, prognostic, and treatment strategies for complex diseases such as clear cell renal cell carcinoma (ccRCC). However, rapid advances of these experimental techniques are yielding vast amounts of heterogeneous data that are becoming progressively more difficult to analyze. Therefore, medical researchers must be provided with increasingly more powerful analysis tools to extract knowledge and develop inferences regarding the roles of genomics, proteomics, and metabolomics in disease process.

Network diagram of cofluctuations for best separators (genes and proteins)

A network showing the cofluctuations of the best separators (genes and proteins). The proximity of the nodes and thickness of the edges represent connection strength; black edges represent positive correlation and yellow edges negative correlations; red nodes represent up-regulation and green nodes represent down-regulation of the genes and proteins in estrogen receptor positive (ER+) vs. estrogen receptor negative (ER-) breast cancer.

For example, the presence of a particular cancer may be signaled by a difference in the protein patterns or gene expression patterns in the body fluids, tissues and sera. Advanced data mining algorithms as well as data management and visualization tools have to be integrated to allow researchers to exploit the progress of experimental techniques and to turn the raw data into discovery and benefits. New data architectures must be developed to address issues such as the rapid integration of analysis of biological data files with epidemiologic profiles of human clinical responses. The search for specific proteins associated with disease is a major challenge. Often the discovery of what seems to be a unique protein biomarker that signals the presence of disease cannot be validated in a clinical setting. Recent evidence also suggests that single biomarkers may not be effective in improving detection, diagnosis and prognosis. This evidence supports approaches that include multiple biomarkers to enhance and improve the reliability of results. Rather than focusing on a single marker, approaches that evaluate multi-marker panels maximize the amount of information that can be extracted from the data.

Example network of differentially expressed genes and proteins linked to cancer

An example network representing the differentially expressed, statistically significant genes and proteins that have been linked to cancer. The triangles represent proteins, circles represent genes. The thickness of the edges (lines) signifies increasing correlation between the nodes. The yellow color represents ccRCC-specific association in current literature. Red border – upregulated, green border – downregulated.

INCOGEN scientists and their collaborators have been involved in the development of statistical pattern recognition and adaptive learning techniques to identify characteristic patterns in gene and protein expression profiles that represent healthy or diseased states. The model systems explored included: clear cell renal cell carcinoma, pediatric Hodgkin's disease, acute myeloid leukemia, and breast cancer. These cancers are remarkably heterogeneous with numerous subtypes and associated prognostic implications. Within each cancer group, treatment may be either intensified or decreased, based on prognosis, in order to improve outcome as well as to avoid side effects of therapy. Accuracy of diagnosis and early stage detection of the disease are life-critical in the fight against cancer.

Laboratory tests have been used to stratify risk and guide medical decision support for decades. In recent years, novel biomarkers and comprehensive biomarker panels have brought these tests into the health care delivery process and are changing the face of medicine. Comprehensive testing provides insight into the individual patient’s pathophysiology, allowing clinicians and other health care professionals to tailor the most appropriate and effective interventions, that may range from lifestyle, to dietary supplementation, to targeted pharmaceutical treatments.

Research Publications →