messina

messina identifies genes with aberrant expression present only in subsets of samples from global expression profiling data (e.g., microarray or proteomic datasets) to detect molecular subtypes in heterogeneous cancers.

Key Features:

• Sensitivity and Specificity: Detects genes with sporadic aberrant expression patterns with an emphasis on high sensitivity and specificity for subset-specific signals.
• Robustness Against Outliers: Implements algorithms for constructing robust single-gene classifiers optimized to handle outliers and unknown sample subgroups.
• Application Across Profiling Data: Operates on global expression profiling data, including microarray and proteomic datasets.
• Identification of Molecular Subtypes: Identifies genes that serve as markers of molecular subtypes, aiding characterization of distinct cancer phenotypes.
• Complementary to Existing Methods: Complements conventional differential expression approaches by detecting genes that are aberrantly expressed only in a subset of samples.

Scientific Applications:

• Clinical biomarker discovery: Facilitates discovery of lead features for development of diagnostic and prognostic clinical tests.
• Molecular subtype characterization and therapeutic stratification: Enables identification of subtype-specific expression patterns to improve understanding of cancer biology and inform targeted therapies, including applications in cancers such as pancreatic cancer.

Methodology:

Constructs robust single-gene classifiers that focus on genes aberrant in sample subsets and are optimized to handle outliers and unknown sample subgroups, applied to global expression profiling data such as microarray and proteomic datasets.