Analysis of array CGH data for cancer studies using fused quantile regression.

Area of Science:

• Genomics
• Cancer Research
• Bioinformatics

Background:

• DNA copy number alterations are crucial in cancer initiation and progression.
• Array-based comparative genomic hybridization (array-CGH) is a high-throughput method for genome-wide scanning of chromosomal aberrations.
• Existing statistical methods for array-CGH data analysis have limitations, including reliance on mean regression and uniform clone spacing, and complex parameter tuning.

Purpose of the Study:

• To develop a more comprehensive and user-friendly statistical method for array-CGH data analysis.
• To enhance the understanding of cancer development by accurately identifying DNA copy number changes.
• To address limitations of existing methods by incorporating physical clone locations and simplifying parameter selection.

Main Methods:

• A fused regularized quantile regression framework was developed for detecting regions of DNA gains and losses.
• Physical locations of clones were incorporated into the model.
• An efficient algorithm was derived to compute the entire solution path for the optimization problem.
• A simple estimate for model complexity was proposed for convenient tuning parameter selection.

Main Results:

• The proposed fused quantile regression model effectively detects regions of gains and losses in array-CGH data.
• Incorporating physical clone locations improves the analysis.
• The developed algorithm efficiently computes the solution path.
• The proposed method for tuning parameter selection simplifies implementation and is demonstrated on three published datasets.

Conclusions:

• The fused quantile regression model offers a more comprehensive and practical approach to array-CGH data analysis.
• This method advances the understanding of cancer genomics by improving the detection of DNA copy number alterations.
• The R code for the method is publicly available, facilitating its adoption in cancer research.