Assessing De Novo transcriptome assembly metrics for consistency and utility

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Transcriptome sequencing and assembly are vital for studying non-model organisms.
• Numerous metrics exist to evaluate transcriptome assemblies, but their accuracy is often unclear.
• Understanding assembly quality is essential for reliable downstream analyses.

Purpose of the Study:

• To evaluate the effectiveness of various metrics in assessing transcriptome assembly quality.
• To determine which metrics accurately reflect assembly completeness and accuracy.
• To guide researchers in selecting appropriate metrics for transcriptome assembly evaluation.

Main Methods:

• Simulated transcriptomic reads from Drosophila melanogaster were generated.
• Assemblies were created using a perfect assembler and a modern transcriptome assembler.
• Read length and sequencing depth were varied to assess metric performance.
• Common, annotation-based, and novel metrics were evaluated.

Main Results:

• Standard metrics like average contig coverage and length were inconsistent unless singletons were included.
• Annotation-based metrics such as contig reciprocal best hit count and unique annotation count proved informative.
• Novel metrics including reverse annotation count, contig collapse factor, and ortholog hit ratio offered unique insights into assembly quality.

Conclusions:

• Few studies have focused on optimizing transcriptome assembly evaluation metrics.
• This research provides a critical review of assembly quality metrics.
• The findings offer practical guidance for producing high-quality transcriptome assemblies.