IsoTree: A New Framework for de novo Transcriptome Assembly from RNA-seq Reads

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• High-throughput sequencing (HTS) of messenger RNA (mRNA) enables deep transcriptome analysis.
• De novo transcriptome assembly from short RNA-sequencing (RNA-seq) reads presents significant algorithmic challenges.

Purpose of the Study:

• To introduce IsoTree, a novel computational framework for reconstructing transcript sequences.
• To enable accurate transcript reconstruction in the absence of a reference genome.

Main Methods:

• IsoTree directly connects RNA-seq reads to construct splicing graphs, bypassing k-mer based graph construction.
• It employs an iterative mixed integer linear programming approach to build an isoform tree from splicing graphs.
• Transcript candidates are identified as paths in the isoform tree and refined using paired-end read information.

Main Results:

• IsoTree demonstrates superior performance compared to leading transcriptome assembly programs in experimental evaluations.
• The framework effectively reconstructs transcript isoforms from complex RNA-seq data.

Conclusions:

• IsoTree provides an effective solution for de novo transcriptome assembly, particularly when reference genomes are unavailable.
• The direct read-to-graph approach offers advantages in transcript reconstruction accuracy and efficiency.