De Novo transcriptome assembly and functional annotation in non-model organisms

Advances in RNA sequencing have established transcriptomics as a key approach for investigating gene expression, regulation, and functional diversity across both model and non-model organisms. In the absence of high-quality reference genomes, de novo transcriptome assembly remains essential, yet its accuracy and reproducibility depend heavily on the choice and integration of computational strategies. This review provides a comprehensive and structured framework for de novo transcriptome reconstruction and functional annotation in non-model systems, integrating current best practices across quality control, assembly, redundancy reduction, and annotation workflows. We systematically compare short-read, long-read, and hybrid assembly approaches, and evaluate multi-layered annotation strategies combining homology-based, domain-based, and ontology-driven methods. We further highlight emerging developments, including long-read transcriptomics, machine learning-assisted annotation, orthology inference, and multi-omics integration, which improve transcriptome completeness and biological interpretability. Key limitations in current workflows are critically examined, including database biases, incomplete annotations, and challenges in isoform resolution, particularly in underrepresented taxa. By consolidating methodological advances with practical recommendations, this review proposes a reproducible and scalable roadmap for transcriptome analysis in non-model organisms. We also discuss future directions, including the development of inclusive reference databases, AI-driven annotation frameworks, and open, containerized infrastructures aligned with FAIR principles to support global transcriptomics research.