Genome-wide repeat landscapes in cancer and cell-free DNA

Area of Science:

• Genomics
• Cancer Biology
• Bioinformatics

Background:

• Genetic alterations in repetitive DNA sequences are common in cancer but difficult to study with standard sequencing.
• Characterizing these repeat elements is crucial for understanding cancer development and progression.

Purpose of the Study:

• To develop a novel computational approach for identifying and analyzing repeat elements in whole-genome sequencing data.
• To investigate tumor-specific changes in repeat element landscapes across various cancer types.
• To explore the potential of repeat element analysis for early cancer detection and origin identification.

Main Methods:

• Developed ARTEMIS (Analysis of RepeaT EleMents in dISease), a de novo kmer finding approach for whole-genome sequencing data.
• Analyzed 1.2 billion kmers from 2837 tissue and plasma samples across 1975 cancer patients.
• Utilized machine learning on genome-wide repeat landscapes and cell-free DNA fragmentation profiles.

Main Results:

• Identified 1280 tumor-specific repeat element types, including 820 novel elements altered in human cancer.
• Found repeat elements enriched in driver gene regions and influenced by structural and epigenetic changes.
• Achieved early-stage lung and liver cancer detection using machine learning on repeat landscapes in cfDNA.
• Demonstrated the ability to noninvasively identify tumor tissue of origin.

Conclusions:

• Widespread alterations in repeat element landscapes are characteristic of human cancers.
• ARTEMIS provides a powerful approach for detecting and characterizing repeat element changes in cancer.
• Repeat element analysis holds significant promise for improving early cancer detection, diagnosis, and disease monitoring.