Unravelling single-cell DNA replication timing dynamics using machine learning reveals heterogeneity in cancer...

Area of Science:

• Genomics
• Computational Biology
• Cancer Research

Background:

• Genomic heterogeneity is often overlooked in single-cell replication timing (scRT) studies.
• Understanding replication timing is crucial for cancer progression research.

Purpose of the Study:

• To develop an efficient machine learning tool (MnM) for disentangling scRT profiles from heterogeneous samples.
• To accurately identify cell replication states and detect genomic heterogeneity.
• To differentiate somatic copy number alterations from DNA replication-associated copy number changes.

Main Methods:

• Utilized single-cell copy number data for imputation and heterogeneity detection.
• Employed machine learning for accurate scRT profile analysis.
• Analyzed over 119,000 human single cells from diverse sources (cell lines, tumors, xenografts).

Main Results:

• Developed MnM, an efficient tool for scRT analysis in heterogeneous samples.
• Successfully separated copy number alterations from replication-driven changes.
• Generated a multi-sample, heterogeneity-resolved scRT atlas from extensive single-cell data.
• Highlighted ubiquitous aneuploidy during tumorigenesis and critical insights into chromosomal aberrations.

Conclusions:

• MnM provides critical insights into chromosomal aberrations and aneuploidy in cancer.
• The scRT atlas is a valuable resource for cancer research, demonstrating scRT's utility in studying replication timing heterogeneity.
• Emphasized the importance of analyzing cancer tissue samples over cell lines for a comprehensive understanding of DNA replication.