Cnngeno: A high-precision deep learning based strategy for the calling of structural variation genotype

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Genotype determination is crucial for understanding organism characteristics.
• Next-generation sequencing accelerates genotype calling, but challenges remain.
• Parametric statistical models struggle with structural variations and coverage fluctuations in genotype calling.

Purpose of the Study:

• To introduce Cnngeno, a novel method for genotype calling of deletions from next-generation sequencing data.
• To leverage convolutional neural networks (CNNs) for improved genotype calling accuracy.

Main Methods:

• Cnngeno converts sequencing data into image representations.
• A convolutional neural network (CNN) is employed for genotype classification from these images.
• A convolutional bootstrapping strategy enhances robustness against noisy labels.

Main Results:

• Cnngeno demonstrates superior precision in genotype calling compared to existing methods.
• The method effectively handles complexities introduced by structural variations and coverage fluctuations.

Conclusions:

• Cnngeno offers a more precise and robust approach to deletion genotype calling.
• The CNN-based method represents a significant advancement in genomic data analysis.