TIGAR2: sensitive and accurate estimation of transcript isoform expression with longer RNA-Seq reads

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• High-throughput RNA sequencing (RNA-Seq) allows precise transcript quantification and identification.
• Advancements in sequencing technologies yield longer reads, necessitating optimized computational methods.
• Existing RNA-Seq quantification tools are not sufficiently adapted for reads exceeding 250 bp.

Purpose of the Study:

• To develop and evaluate TIGAR2, a novel statistical method for transcript isoform quantification.
• To address the challenge of quantifying gene expression from both fixed and variable-length RNA-Seq reads, particularly longer ones.

Main Methods:

• TIGAR2 models sequencer errors (substitutions, deletions, insertions) using gapped alignments.
• Integrates sensitive read-aligners like Bowtie2 and BWA-MEM.
• Employs a heuristic algorithm within variational Bayesian inference for efficient computation.

Main Results:

• TIGAR2 demonstrates high sensitivity and accuracy in quantifying transcript isoform abundances.
• Performance evaluation on simulated and real human RNA-Seq data confirms TIGAR2's efficacy.
• TIGAR2 outperforms existing methods for various read lengths, including fixed (100-1000 bp) and variable-length reads.

Conclusions:

• TIGAR2 is a robust and accurate tool for transcript isoform quantification from RNA-Seq data.
• The method shows superior performance, especially for longer reads (>250 bp), enhancing gene expression analysis.