Clustering Single-Cell Expression Data Using Random Forest Graphs.

Area of Science:

• Computational Biology
• Genomics
• Biotechnology

Background:

• Complex tissues comprise diverse cell types, necessitating cell-type-specific analysis.
• Advancements in single-cell cytometry yield high-throughput biological data.
• Machine learning is crucial for interpreting complex cellular landscapes.

Purpose of the Study:

• To propose a novel machine learning technique for cell-type profiling in intricate tissues.
• To leverage single-cell cytometry data for enhanced cellular landscape analysis.

Main Methods:

• Developed a random-forest-based single-cell profiling method.
• Utilized random forests to model cell marker dependencies and cellular populations.
• Employed the cell network concept to identify distinct cell types within tissues.

Main Results:

• The proposed technique demonstrates promising performance in profiling complex tissues.
• Experimental results on public datasets show high accuracy in cell population extraction.
• The method effectively captures cell marker interdependencies for robust cell identification.

Conclusions:

• Random-forest-based single-cell profiling is an accurate method for dissecting complex tissue cellularity.
• The cell network concept aids in discovering and characterizing cell populations.
• This technique offers valuable insights into the cellular composition of intricate biological tissues.