Challenges and advances for transcriptome assembly in non-model species.

Area of Science:

• Bioinformatics
• Genomics
• Computational Biology

Background:

• Transcriptome assembly for non-model organisms typically uses de novo or genome-guided approaches.
• Genome-guided assembly faces limitations with highly divergent reference genomes, common in non-model species.
• Traditional mapping methods struggle with >15% genetic divergence.

Purpose of the Study:

• To evaluate blastn as a superior read assignment method for transcriptome-guided assembly in non-model organisms with high genetic divergence.
• To compare the performance of blastn-guided assembly against de novo assembly alone and in combination with a reference transcriptome.
• To provide guidelines for data processing in comparative transcriptomics and gene expression analyses.

Main Methods:

• Simulated high-throughput transcriptome reads with varying lengths and genetic divergence (0-30%).
• Utilized blastn for read assignment in transcriptome-guided assembly.
• Compared blastn-guided assembly with de novo assembly using simulated and empirical data from fish (Parachondrostoma toxostoma) and oak (Quercus pubescens) species.

Main Results:

• Blastn-guided assembly recovered 94.8% of genes at 0% divergence and 92.6% at 30% divergence, irrespective of read length.
• Read assignment rate correlated negatively with divergence level and read length.
• Blastn-guided assembly consistently outperformed de novo assembly, especially at higher divergence levels, and improved contiguity and completeness metrics in empirical data.

Conclusions:

• Blastn-guided transcriptome assembly is a robust method for gene identification in non-model organisms, outperforming de novo assembly, particularly when genetic divergence is high.
• Combining de novo assembly with blastn-guided assembly mitigates biases associated with relying solely on a related reference.
• The findings offer practical guidance for processing transcriptomic data to minimize inferential bias in comparative analyses.