Piecewise polynomial representations of genomic tracks

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Genomic data involves mapping measured values to chromosomal coordinates.
• These associations can be modeled as one-dimensional functions.
• Existing methods may have limitations in analyzing complex genomic signals.

Purpose of the Study:

• To present a general framework for creating piecewise polynomial representations of genome-scale signals.
• To demonstrate the utility of this framework in analyzing various genomic data types.
• To offer improved methods for copy-number analysis and transcription level assessment.

Main Methods:

• Developing a general framework for piecewise polynomial modeling of genomic signals.
• Applying piecewise constant segmentation for copy-number analysis in array and sequencing data.
• Utilizing higher-order polynomial curves to analyze trends and discontinuities in RNA-seq data.
• Employing piecewise linear functions for diagnosing alignment quality at splice sites.

Main Results:

• Piecewise constant segmentation within the framework offers advantages for both array and DNA sequencing data.
• Higher-order polynomials can effectively detect trends and discontinuities in RNA-seq data.
• Piecewise linear functions provide a method to diagnose and quantify alignment quality at exon borders.

Conclusions:

• The presented piecewise polynomial framework offers a versatile approach for analyzing diverse genome-scale signals.
• This method provides advantages over existing techniques for copy-number analysis.
• The framework facilitates the detection of trends and assessment of alignment quality in genomic data.