The impacts of read length and transcriptome complexity for de novo assembly: a simulation study

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• RNA sequencing (RNA-seq) enables transcriptome assembly, crucial for understanding gene expression, especially in non-model organisms.
• Pre-assembly steps like sequencing depth and error correction are vital, but the impact of read length and assembler performance across varying transcriptome complexities remain understudied.

Purpose of the Study:

• To investigate the impact of read length on transcriptome assembly quality.
• To evaluate the performance of current de novo assembly tools with increasing transcriptome complexity and alternative splicing events.

Main Methods:

• Utilized two high-performing assemblers, Velvet/Oases and Trinity.
• Employed simulated datasets from human, mouse, and S. cerevisiae with varying complexities.
• Analyzed the effect of read length and transcriptome complexity on assembly accuracy.

Main Results:

• Read length has a diminishing impact on assembly quality once a specific organism-dependent threshold is met.
• Assembly quality significantly degrades as transcriptome complexity increases.
• Existing de novo assemblers struggle and often produce corrupted assemblies when faced with high numbers of alternative splicing events.

Conclusions:

• Optimizing read length beyond a certain point offers limited benefits for transcriptome assembly.
• Transcriptome complexity and alternative splicing present significant challenges for current de novo assembly tools.
• Further development of assemblers is needed to accurately reconstruct transcriptomes with complex gene structures.