The paper introduces Oases, a software package designed for de novo transcriptome assembly from RNA-seq data without a reference genome. Oases addresses the challenges of uneven expression levels and alternative splicing by using multiple hash lengths, dynamic noise filtering, robust resolution of alternative splicing, and efficient merging of assemblies. The authors tested Oases on human and mouse RNA-seq data, demonstrating its superior performance compared to other de novo assemblers like transAbYSS and Trinity. Oases was found to produce longer assemblies and achieve higher sensitivity, especially at high expression levels. The study also highlights the importance of exploring different k-mer lengths to adapt to varying conditions and the benefits of merging assemblies from multiple algorithms. Overall, Oases provides a robust pipeline for assembling full-length transcripts from RNA-seq data, particularly in the absence of a reference genome.The paper introduces Oases, a software package designed for de novo transcriptome assembly from RNA-seq data without a reference genome. Oases addresses the challenges of uneven expression levels and alternative splicing by using multiple hash lengths, dynamic noise filtering, robust resolution of alternative splicing, and efficient merging of assemblies. The authors tested Oases on human and mouse RNA-seq data, demonstrating its superior performance compared to other de novo assemblers like transAbYSS and Trinity. Oases was found to produce longer assemblies and achieve higher sensitivity, especially at high expression levels. The study also highlights the importance of exploring different k-mer lengths to adapt to varying conditions and the benefits of merging assemblies from multiple algorithms. Overall, Oases provides a robust pipeline for assembling full-length transcripts from RNA-seq data, particularly in the absence of a reference genome.