A Study of Cell-free DNA Fragmentation Pattern and Its Application in DNA Sample Type Classification

Area of Science:

• Genomics
• Bioinformatics
• Molecular Diagnostics

Background:

• Plasma cell-free DNA (cfDNA) exhibits distinct fragmentation patterns.
• These patterns influence base composition curves in sequencing data.
• Distinguishing cfDNA from other DNA sources is crucial for accurate analysis.

Purpose of the Study:

• To investigate unique fragmentation patterns in cfDNA.
• To develop a method for identifying cfDNA based on sequencing data.
• To assess the accuracy of cfDNA classification models.

Main Methods:

• Analysis of 3189 FastQ files, including cfDNA, genomic DNA, and FFPE tumor DNA.
• Examination of base content curves in the initial sequencing cycles.
• Development and training of classification models (KNN, Random Forest, SVM) using pattern recognition.

Main Results:

• cfDNA exhibits stable, unique base content patterns in early sequencing cycles.
• Classification models achieved average accuracies exceeding 98%.
• Random Forest classifier yielded the highest accuracy at 99.89% (σ = 0.00068).

Conclusions:

• Sequencing data's initial base content patterns reliably identify cfDNA.
• Computational models can accurately distinguish cfDNA from other DNA samples.
• The CfdnaPattern tool facilitates cfDNA identification from sequencing data.