MiSTIC

MiSTIC visualizes and analyzes genome-wide transcriptome data to identify gene correlation structures and infer tumor subgroups for interpretation of cancer subtypes, prognostic signatures, transcription factor roles, and biomarker interactions.

Key Features:

• Interactive Visualization: Visualizes and compares gene correlation structures across different transcriptome datasets to reveal co-expression patterns.
• Molecular Analysis: Analyzes molecular causes underlying co-variations in gene expression among cancer samples to interpret the biological significance of subtypes.
• Clinical Annotation Integration: Integrates clinical annotations of tumor sets based on combined expression profiles of selected biomarkers.
• Transcription Factor Role Analysis: Elucidates roles of specific transcription factors in breast cancer subtype specification.
• Prognostic Signature Comparison: Compares prognostic signatures and their targeted aspects of tumor heterogeneity.
• Biomarker Interaction Highlighting: Highlights interactions between biomarkers, including applications in acute myeloid leukemia (AML).
• Unsupervised Classification: Performs non-supervised classification of tumours using genome-wide transcriptome profiling to identify distinct sub-groups with unique gene expression features.

Scientific Applications:

• Transcriptome-level analysis: Integrates visualization, molecular analysis, and clinical annotation to analyze genome-wide transcriptome data.
• Cancer subtype discovery: Uncovers biological underpinnings of cancer subtypes, including transcription factor-driven specification in breast cancer.
• Prognosis and biomarker evaluation: Compares prognostic signatures and links biomarker expression profiles to clinical outcomes for prognosis and treatment implications.
• Biomarker interaction studies in AML: Identifies and highlights biomarker interactions in acute myeloid leukemia to inform disease-mechanism investigations.

Methodology:

MiSTIC employs a minimum spanning tree approach to infer clustering within transcriptome datasets and performs non-supervised classification using genome-wide transcriptome profiling.