Premnas

Premnas estimates subpopulation composition from bulk L1000 gene expression profiles to reveal intraclonal heterogeneity and perturbation-induced compositional changes in Connectivity Map (CMap) datasets.

Key Features:

• Subpopulation Estimation: Premnas estimates the abundance of undetermined subpopulations within L1000 gene expression profiles from CMap datasets and derives an ad hoc subpopulation representation via archetypal analysis of well-normalized single-cell RNA-seq datasets.
• Digital Cytometry: Premnas applies digital cytometry to deconvolute bulk expression profiles into subpopulation compositions and quantify dynamic changes upon perturbation.
• Enhanced Interpretability and Reproducibility: Premnas recovers subpopulation-level responses to perturbagens to improve interpretability and reproducibility of CMap dataset connections and to reveal cellular responses obscured by bulk profiling.
• Drug Discovery Applications: Premnas identifies potential drug cocktails by analyzing subpopulation dynamics and links drug-resistant subpopulations in CMap L1000 profiles to known resistant clones such as pre-adapted (PA) cells.

Scientific Applications:

• Understanding Therapeutic Pathways: Premnas reveals intraclonal and intratumoral heterogeneity and expands perturbation datasets like CMap to connect cellular signatures across treatments.
• Drug Resistance Research: Premnas identifies and characterizes drug-resistant subpopulations to inform strategies for overcoming therapeutic resistance.

Methodology:

Archetypal analysis is applied to well-normalized single-cell RNA-seq datasets to learn ad hoc subpopulation representations, which are used with digital cytometry to deconvolute bulk L1000 gene expression profiles from CMap and estimate subpopulation abundance changes upon perturbation.