puma 3.0: improved uncertainty propagation methods for gene and transcript expression analysis.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Microarrays are established tools for genome-scale gene expression profiling.
• The Bioconductor package puma offers uncertainty analysis for Affymetrix GeneChip data.
• Exon microarray technology is increasingly used for isoform-level expression analysis, but current tools like puma lack support.

Purpose of the Study:

• To extend the puma package with methods for exon array data analysis.
• To improve existing downstream analysis methods for gene and isoform expression.
• To enhance the stability and efficiency of expression quantification and analysis.

Main Methods:

• Developed a new gamma model for exon arrays to estimate isoform and gene expression with uncertainty.
• Implemented a variant of the probe-level analysis for more stable gene expression estimates from 3' GeneChip data.
• Introduced a computationally efficient differential expression detection method and a robust clustering method considering replicate information.

Main Results:

• The extended puma package now supports both Affymetrix 3' GeneChips and Exon arrays.
• Isoform and gene expression levels with associated uncertainties are calculated using a novel gamma model for exon arrays.
• Improved methods provide more stable expression estimates, efficient differential expression detection, and robust clustering.

Conclusions:

• The enhanced puma package provides comprehensive gene and isoform expression analysis for both 3' and Exon arrays.
• Uncertainty propagation is integrated into downstream analyses, improving efficiency and applicability.
• The downstream methods can be adapted for other platforms like RNA-Seq when uncertainty data is available.