Related Experiment Videos
Analysis and visualization of functional relationships between RNA expression and clinical annotation using PathlinX
1Bioinformatics, Xpogen Incorporated, Cambridge, MA, USA. scarter@xpogen.com
Proceedings. AMIA Symposium
|December 5, 2002
Summary
PathlinX, an unsupervised analysis tool, reveals significant gene expression patterns in lung cancer data. This method helps researchers understand complex relationships between patient data and tumor characteristics.
Area of Science:
- Oncology
- Bioinformatics
- Genomics
Background:
- Lung carcinoma gene expression data is complex and often heterogeneous.
- Understanding relationships between gene expression and clinical parameters is crucial for cancer research.
Purpose of the Study:
- To demonstrate PathlinX's capability for unsupervised analysis of heterogeneous biological data.
- To identify significant associations within lung carcinoma gene expression and clinical datasets.
Main Methods:
- Analysis of a publicly available dataset of 105 lung carcinomas with gene-expression and clinical data.
- Empirical evaluation of metrics to distinguish biological signal from noise using data permutation.
- Generation of PathlinX networks by grouping significant associations via transitive closure.
Main Results:
- Identification of significant associations between gene expression and clinical parameters in lung carcinomas.
- Demonstration of PathlinX's ability to highlight key biological features in large datasets.
- Establishment of significance thresholds based on permuted data analysis.
Conclusions:
- PathlinX provides an effective unsupervised approach for exploring complex biological datasets.
- The technique facilitates rapid intuition into significant relationships within heterogeneous data.
- PathlinX networks can effectively highlight critical biological insights in cancer research.