SRIQ

SRIQ performs unsupervised clustering of gene expression data to identify biologically meaningful tumor subtypes while avoiding pre-specification of the number of clusters and reducing subjectivity via feature pre-selection.

Key Features:

• Integration of Random Forest Concepts: Incorporates machine learning techniques from random forests to enhance clustering and pattern identification in complex gene expression datasets.
• Quality Threshold and k-Nearest Neighbor Clustering: Combines quality threshold methods with k-nearest neighbor clustering to identify core clusters of highly correlated observations and expand them into larger clusters.
• Automatic Evaluation of Stability: Autonomously evaluates clustering solution stability without requiring the user to pre-specify the number of cluster solutions to assess.
• Feature Pre-selection: Employs feature pre-selection to reduce subjectivity in input feature choice for clustering.
• Technical Reproducibility: Demonstrated reproducibility on 434 lung adenocarcinomas profiled by RNA sequencing.
• Comprehensive Pipeline Components: Includes data pre-processing, differential expression analysis, and pathway analysis as part of the analytical workflow.

Scientific Applications:

• Molecular subtype discovery: Identification of new molecular subtypes of tumors from gene expression data in cancer research.
• Refinement of transcriptional subtypes: Refinement of existing transcriptional subtypes defined by consensus clustering to provide more nuanced subclassification.
• Disease classification and clinical insight: Support for improved disease classification and for informing personalized treatment strategies based on transcriptional subtypes.
• Application to lung adenocarcinoma: Empirical application to RNA sequencing data from lung adenocarcinomas to define reproducible tumor subsets.

Methodology:

Uses feature pre-selection, integration of random forest concepts, quality threshold and k-nearest neighbor clustering, automatic evaluation of cluster stability, and includes data pre-processing, differential expression analysis, and pathway analysis.