PenDA

PenDA infers gene-level deregulation in individual samples by analyzing the local ordering of gene expression to enable personalized differential analysis against a reference dataset.

Key Features:

• Personalized differential analysis: PenDA performs differential analysis at the single-sample scale by leveraging local ordering of gene expression within individual cases to infer deregulation relative to a reference dataset.
• High specificity and sensitivity: In realistic simulations using RNA-seq tumor data, PenDA demonstrated performance over existing methods and maintained robustness to normalization effects.
• Cancer-type-specific insights: Application to lung cancer cohorts revealed that deregulated genes show commitment toward either up-regulation or down-regulation specific to the cancer type.

Scientific Applications:

• Molecular histology definition: Individual deregulation profiles from PenDA were used to define new molecular histologies for lung adenocarcinoma that strongly correlate with patient survival.
• Biomarker identification: PenDA identified 37 biomarkers up-regulated in tumors and associated with poor prognosis, validated across two independent cohorts.
• Therapeutic target discovery: Personalized deregulation patterns extracted by PenDA aid in identifying potential therapeutic targets for personalized diagnosis and treatment strategies.

Methodology:

PenDA uses local ordering of gene expression within individual samples and comparison against reference datasets to identify gene deregulations, and it is robust to normalization effects.