Mixture models reveal multiple positional bias types in RNA-Seq data and lead to accurate transcript concentration...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA sequencing (RNA-Seq) is crucial for gene expression analysis.
• Positional fragment bias negatively impacts transcript quantification accuracy.
• Existing methods for bias correction have limitations.

Purpose of the Study:

• Introduce Mix2, a novel method for transcript quantification.
• Address and neutralize positional fragment bias in RNA-Seq data.
• Evaluate Mix2's performance against state-of-the-art methods.

Main Methods:

• Mix2 utilizes a mixture of probability distributions to model bias.
• Expectation Maximization algorithm trains Mix2 parameters for abundance and bias estimation.
• Comparison with Cufflinks, RSEM, eXpress, and PennSeq on synthetic and real RNA-Seq data.

Main Results:

• Mix2 demonstrates superior accuracy and bias estimation on synthetic data.
• On MAQC data, Mix2 shows improved correlation with qPCR and enhanced reproducibility.
• Mix2 achieves more accurate differential expression detection and reveals novel biases.
• On SEQC data, Mix2 provides higher consistency in concentration ratios and improved inter-laboratory repeatability.

Conclusions:

• Mix2 effectively mitigates positional fragment bias in RNA-Seq.
• The method offers significant improvements in accuracy, reproducibility, and differential expression analysis.
• Mix2 provides a more reliable approach for transcript quantification compared to existing tools.