Effective Clustering for Single Cell Sequencing Cancer Data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single cell sequencing (SCS) is crucial for inferring tumor phylogenies by analyzing accumulated mutations.
• High false negative and missing value rates in SCS data create computational challenges for tree inference.
• Reducing SCS data size is a key strategy to overcome these limitations.

Purpose of the Study:

• To introduce a novel clustering procedure, celluloid, for SCS data.
• To improve the efficiency and accuracy of phylogenetic inference from SCS data.
• To demonstrate the practical utility of celluloid in analyzing large-scale SCS datasets.

Main Methods:

• Developed celluloid, a clustering procedure for categorical vector/matrix data representing SCS instances.
• Applied celluloid to cluster mutations within SCS data.
• Integrated celluloid into a computational pipeline for phylogenetic inference.

Main Results:

• Celluloid accurately clusters mutations, avoiding the pairing of unrelated mutations.
• Phylogenetic inference downstream of celluloid clustering yields accurate results.
• The celluloid pipeline significantly reduces runtime and increases the scalability of SCS data analysis.

Conclusions:

• Celluloid effectively addresses the challenge of high dimensionality and noise in SCS data.
• The celluloid clustering approach enhances the accuracy and efficiency of tumor evolution inference.
• This method expands the practical applicability of SCS data analysis for complex biological systems.