Benchmark analysis of algorithms for determining and quantifying full-length mRNA splice forms from RNA-seq data

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• RNA sequencing (RNA-Seq) offers advantages over microarrays for gene expression analysis.
• Accurate identification and quantification of full-length splice forms using RNA-Seq is a key goal.
• Existing informatics packages struggle with short reads, sequencing errors, and polymorphisms, necessitating algorithm evaluation.

Purpose of the Study:

• To evaluate the accuracy of existing bioinformatics algorithms for RNA sequencing data analysis.
• To identify the best-performing algorithms for gene expression analysis.
• To address the lack of independent and unbiased benchmarking studies for RNA-Seq analysis tools.

Main Methods:

• Utilized both simulated and experimental benchmark datasets for evaluation.
• Assessed algorithm performance under various complicating factors including multiple splice forms, polymorphisms, and sequencing/alignment errors.
• Provided access to simulated datasets and supporting information for reproducibility.

Main Results:

• Most tested algorithms demonstrated inaccuracy, even with idealized RNA sequencing data.
• No algorithm achieved high accuracy when complex factors like multiple splice forms, polymorphisms, and various errors were present.
• Identified significant limitations in current computational methods for analyzing RNA-Seq data.

Conclusions:

• Current RNA sequencing analysis algorithms are largely inaccurate, especially with real-world data complexities.
• There is an urgent need for the development of more robust and accurate bioinformatics algorithms.
• Further research and development are critical to fully leverage the potential of RNA-Seq for gene expression studies.