Gene-expression memory-based prediction of cell lineages from scRNA-seq datasets

Area of Science:

• Computational Biology
• Genomics
• Developmental Biology

Background:

• Lineage tracing is crucial for understanding cell differentiation but is technically challenging and often lacks temporal resolution.
• Most single-cell RNA sequencing (scRNA-seq) datasets do not contain inherent lineage information.
• Existing methods struggle to identify small to medium-sized cell lineages.

Purpose of the Study:

• To introduce Gene Expression Memory-based Lineage Inference (GEMLI), a computational tool for inferring cell lineages solely from scRNA-seq data.
• To enable the study of heritable gene expression, cell fate decisions, and multicellular structure reconstruction.
• To identify novel gene expression changes associated with cancer invasiveness.

Main Methods:

• GEMLI utilizes scRNA-seq data to computationally reconstruct cellular lineage trees.
• The method analyzes gene expression patterns to infer lineage relationships.
• Application to human breast cancer biopsies to identify early invasive changes.

Main Results:

• GEMLI successfully identifies small to medium-sized cell lineages from scRNA-seq data.
• The tool can discriminate between symmetric and asymmetric cell fate decisions.
• Novel gene expression alterations at the onset of cancer invasiveness were discovered in human breast cancer samples.

Conclusions:

• GEMLI provides a robust computational approach to infer cell lineages without experimental lineage tracing.
• The tool facilitates the study of cell lineage dynamics in various physiological and pathological contexts.
• GEMLI offers a universal applicability for investigating the role of cell lineages in vivo.