Estimating the differentiation potential and plasticity of cancer cells using statistical mechanics

• Cell differentiation is a crucial property of both normal and cancerous cells, that is driven by complex underlying processes. A number of computational methods can score the differentiation potential of individual cells based on their RNA expression. However, we lack a unifying model to explain how differentiation arises from underlying gene regulation and external perturbations. Here, we show that an adaptation of the Ising model, commonly used in statistical mechanics, can bridge this gap, thereby offering a way to identify normal and cancer stem cells. Our new model states that every cell updates its gene expression pattern according to a Boltzmann distribution, influenced by the gene-gene network and an external perturbation field. We first show that this model can be fitted to scRNAseq data sets. We apply the model to a range of data sets to demonstrate its efficacy in separating cells with varying differentiation potential and creating a pseudo-temporal ordering of cells in a GBM data set. Additionally, we explore other aspects of the model to identify known chromosomal aberrations of GBM from single cells and predict therapeutic interventions. This framework has potential applications in many cancer types and can be used to identify CSCs and measure differentiation potential without relying on stemness signatures or marker genes. 

Ämnesord

NATURVETENSKAP  -- Data- och informationsvetenskap -- Bioinformatik (hsv//swe)

NATURAL SCIENCES  -- Computer and Information Sciences -- Bioinformatics (hsv//eng)

Nyckelord

differentiation potential

plasticity

single-cell profiling

the ising model

computational modeling

gene perturbations

Oncology

Onkologi

Oncology

Onkologi

Oncology

Onkologi

Publikations- och innehållstyp

vet (ämneskategori)

ovr (ämneskategori)