Identifying disease-associated copy number variations by a doubly penalized regression model

Area of Science:

• Genetics
• Genomics
• Bioinformatics

Background:

• Copy number variations (CNVs) are implicated in numerous diseases.
• Analyzing CNV-disease associations is challenging due to CNV irregularity and sparsity.
• Existing methods often use a less powerful two-stage approach, leading to information loss.

Purpose of the Study:

• To develop a novel, integrated method for identifying disease-associated common CNVs.
• To improve the power of detecting CNVs linked to specific disease traits.
• To address limitations of current two-stage analytical procedures.

Main Methods:

• A single, coherent model combining CNV identification and association testing.
• Utilizing a double penalty model to capture associations with both DNA intensities and disease traits.
• Validation through simulated datasets and a real-world ovarian cancer genome dataset.

Main Results:

• The proposed method effectively integrates CNV identification and association testing.
• Demonstrated improved power in detecting disease-associated CNVs compared to traditional methods.
• Successfully applied to identify CNVs associated with platinum resistance in ovarian cancer.

Conclusions:

• The integrated, double penalty model offers a more powerful approach for CNV-disease association studies.
• This method provides a robust framework for analyzing complex genomic variations in disease.
• Highlights the potential of integrated analysis in uncovering novel disease-related genetic factors.