PyEvoCell: an LLM-augmented single-cell trajectory analysis dashboard

Area of Science:

• Computational Biology
• Bioinformatics
• Single-cell Genomics

Background:

• Trajectory inference in single-cell studies is complex.
• Interpreting cell lineages and downstream analysis requires expertise.
• Existing methods lack integrated interpretation support.

Purpose of the Study:

• To develop an advanced dashboard for single-cell trajectory analysis.
• To leverage large language model (LLM) capabilities for enhanced interpretation.
• To provide a tool for hypothesis generation and validation in single-cell studies.

Main Methods:

• Developed PyEvoCell, a dashboard integrating LLM functionalities.
• Applied LLM to trajectory inference outputs (e.g., Monocle3) for lineage suggestion.
• Integrated differential expression, functional analysis, and a veracity filter powered by LLM and PubMed.

Main Results:

• PyEvoCell suggests biologically relevant cell lineages.
• LLM interprets differential expression and functional analysis results.
• Veracity filter validates hypotheses by citing PubMed articles.

Conclusions:

• PyEvoCell facilitates the analysis and interpretation of single-cell trajectories.
• LLM integration enhances biological insight and hypothesis validation.
• The tool streamlines complex single-cell data analysis workflows.