Mapping Phenotypic Plasticity upon the Cancer Cell State Landscape Using Manifold Learning.

Area of Science:

• Cancer Biology
• Computational Biology
• Genomics

Background:

• Phenotypic plasticity, driven by nongenetic mechanisms, significantly contributes to tumor heterogeneity, metastasis, and therapy resistance.
• Cancer cells exist on a spectrum of dynamic states, forming a cell state landscape.
• Existing technologies allow for systematic recording of molecular mechanisms at single-cell resolution.

Purpose of the Study:

• To model cancer cell state dynamics and understand the cell state landscape.
• To explore the potential of manifold learning techniques in modeling these dynamics.
• To propose novel therapeutic strategies targeting phenotypic plasticity.

Main Methods:

• Utilizing manifold learning techniques to model cell state dynamics.
• Analyzing single-cell resolution data to capture molecular mechanisms.
• Defining phenotypic plasticity as a framework for therapeutic development.

Main Results:

• Manifold learning effectively models cell state dynamics, mimicking the cell state landscape.
• Nongenetic mechanisms of phenotypic plasticity are key drivers of cancer progression.
• Identification of potential vulnerabilities for therapeutic exploitation.

Conclusions:

• Phenotypic plasticity is a critical factor in cancer heterogeneity, metastasis, and therapy resistance.
• Manifold learning offers a powerful computational tool for studying cell state dynamics.
• "State-gating" therapies targeting phenotypic plasticity show promise in limiting cancer progression and resistance.