SIGMA2: A system for the integrative genomic multi-dimensional analysis of cancer genomes, epigenomes, and transcriptomes

Raj Chari, Bradley P. Coe, Craig Wedseltoft, Marie Benetti, Ian M. Wilson, Emily A. Vucic, Calum MacAulay, Raymond T. Ng, Wan L. LamBMC Bioinformatics 2008 · DOI

TL;DR

SIGMA2 is a novel software system designed to integrate and analyze multi-dimensional 'omics data (genomics, epigenomics, and transcriptomics) from cancer samples. It addresses the challenge of understanding complex cancer mechanisms by allowing simultaneous visualization and analysis across these different biological layers, facilitating the discovery of genes altered at multiple levels.

What Problem Does This Paper Address?

Cancer is a complex disease driven by disruptions at multiple biological levels, including genetic alterations (like copy number changes and loss of heterozygosity), epigenetic modifications (such as DNA methylation), and their downstream effects on gene expression (transcriptomics). While high-throughput technologies allow us to assay these different "omics" dimensions with unprecedented resolution, existing software often focuses on analyzing only one or two dimensions at a time. This limitation hinders a comprehensive understanding of how these alterations interact and contribute to cancer development. Researchers need tools that can integrate and analyze data from genomics, epigenomics, and transcriptomics simultaneously to uncover the intricate mechanisms underlying cancer.

Methods

SIGMA2 is implemented in Java and leverages the statistical power of R for analysis, with data stored and queried in a MySQL database. Key methodological aspects include:

  • Interval-based Integration: SIGMA2 treats all data within the context of genomic position, using human genome builds from UCSC. It employs an interval-based approach to sample data across different array platforms and assays, merging information by averaging measurements within fixed genomic intervals. This allows for the aggregation of data from diverse platforms measuring the same biological attribute.
  • Multi-dimensional Visualization and Analysis: The system allows for simultaneous visualization and analysis of data across genomics, epigenomics, and transcriptomics. This enables users to explore how alterations in one dimension (e.g., copy number gain) correlate with changes in another (e.g., gene expression).
  • Consensus Calling: For segmentation analysis of array CGH data, SIGMA2 can take a consensus from multiple algorithms, allowing users to specify logic (AND/OR) and a minimum percentage of agreement for detected changes.
  • Statistical Comparisons: SIGMA2 supports various statistical analyses, including two-group comparisons using Fisher's Exact Test for identifying statistically different patterns of alteration between groups, and Mann-Whitney U tests for comparing gene expression between samples with and without specific genomic alterations.
  • Integration with Biological Databases: The system links to external biological databases (PubMed, OMIM, NCBI Gene, UCSC Genome Browser, GEO Profiles, Database of Genomic Variants) to facilitate further interrogation of identified genes and regions.

Datasets

The paper describes the use of various types of biological data, including:

  • Genomic Data: Array CGH (copy number), SNP arrays (LOH), microsatellite markers (LOH).
  • Epigenomic Data: MeDIP (DNA methylation), ChIP-on-chip (histone modification).
  • Transcriptomic Data: Microarrays, SAGE (gene expression).

The paper uses examples from breast cancer cell lines (HCC2218) and lung cancer cell lines (NSCLC and SCLC) to demonstrate the functionality of SIGMA2. It does not introduce new benchmark datasets but rather showcases the integration of existing data types.

Key Insights and Ideas

  • The Power of Integration: The core insight is that understanding cancer requires looking beyond single 'omics dimensions. SIGMA2 demonstrates how integrating genomic, epigenomic, and transcriptomic data can reveal complex, multi-layered alterations that are missed by single-dimension analyses.
  • Uncovering "Two-Hit" and "Or" Events: The system can identify genes affected by multiple mechanisms simultaneously ("two-hit" events, e.g., copy number change and methylation alteration) or genes where alteration in any of several dimensions can lead to a similar outcome ("or" events). This is crucial for understanding compensatory mechanisms and diverse pathways leading to a common phenotype.
  • Bridging Genotype and Phenotype: By linking genomic alterations to gene expression, SIGMA2 helps bridge the gap between genotype and phenotype, allowing researchers to hypothesize how specific genetic or epigenetic changes might drive observable cellular or organismal traits.
  • Facilitating Systems Biology: SIGMA2 provides a framework for high-throughput systems biology analysis of cancer, enabling researchers to move from individual gene studies to a more holistic understanding of cellular networks and pathways.

Key Results

The paper showcases SIGMA2's capabilities through several examples:

  • Breast Cancer Cell Line Analysis: In the HCC2218 breast cancer cell line, SIGMA2 integrated genomic, epigenomic, and transcriptomic data to reveal concurrent amplification of the ERBB2 gene, coinciding with allelic imbalance, loss of methylation, and a drastic increase in gene expression. This highlights how multiple alterations can converge on a single gene critical for cancer development.
  • Lung Cancer Cell Line Comparison: A two-group comparison of NSCLC and SCLC cell lines demonstrated how SIGMA2 could identify regions of differential copy number gain (e.g., on chromosome 7p) and then correlate this with gene expression data, showing significantly higher EGFR expression in NSCLC, consistent with the observed copy number gain.
  • Heterogeneous Array Analysis: The system successfully integrated data from different array platforms (Affymetrix SNP 500K, Agilent 244K CGH, and whole genome BAC array) for a single sample, generating a frequency histogram of common areas of gain and loss across these diverse data sources.

Limitations and Future Work

  • Future Requirement for Sequence Data: The authors acknowledge that with the movement from array to sequence-based technologies, the ability to assimilate sequence data with 'omics data sets will become a future requirement for software packages like SIGMA2.
  • Commercial Use Licensing: While free for academic and research use, commercial users need to contact the authors for licensing.
  • Genome Build Updates: The availability of new genome builds will require updates to the system to maintain compatibility.

The paper suggests that SIGMA2 provides a robust framework that can be extended to incorporate future technological advancements and data types, paving the way for more comprehensive cancer research.

Summary updated

BibTeX
@article{chari2008sigma,
  author = {Raj Chari and Bradley P. Coe and Craig Wedseltoft and Marie Benetti and Ian M. Wilson and Emily A. Vucic and Calum MacAulay and Raymond T. Ng and Wan L. Lam},
  title = {{SIGMA2: A system for the integrative genomic multi-dimensional analysis of cancer genomes, epigenomes, and transcriptomes}},
  journal = {BMC Bioinformatics},
  year = {2008},
  doi = {10.1186/1471-2105-9-422},
}
Figure