This study investigates the impact of sequencing depth on RNA-seq data analysis, focusing on the detection and identification of differentially expressed transcripts. The authors analyze how sequencing depth affects transcript biotype, length, expression level, and fold-change, using three human RNA-seq datasets with varying depths. They propose a novel nonparametric method called NOISEq, which models the noise distribution from the actual data, making it more robust to sequencing depth and better at controlling false positives. The results show that existing methods are highly dependent on sequencing depth, leading to a significant number of false positives, especially in low-expression ranges. In contrast, NOISEq maintains a stable and low false-discovery rate (FDR) even with higher sequencing depths, demonstrating its effectiveness in differential expression analysis. The study also highlights the importance of balanced sequencing depths between samples and the need for careful consideration of transcript length and expression levels when interpreting differential expression results.This study investigates the impact of sequencing depth on RNA-seq data analysis, focusing on the detection and identification of differentially expressed transcripts. The authors analyze how sequencing depth affects transcript biotype, length, expression level, and fold-change, using three human RNA-seq datasets with varying depths. They propose a novel nonparametric method called NOISEq, which models the noise distribution from the actual data, making it more robust to sequencing depth and better at controlling false positives. The results show that existing methods are highly dependent on sequencing depth, leading to a significant number of false positives, especially in low-expression ranges. In contrast, NOISEq maintains a stable and low false-discovery rate (FDR) even with higher sequencing depths, demonstrating its effectiveness in differential expression analysis. The study also highlights the importance of balanced sequencing depths between samples and the need for careful consideration of transcript length and expression levels when interpreting differential expression results.