The paper "Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences" by Charlotte Soneson, Michael I. Love, and Mark D. Robinson discusses the advantages of using gene-level abundance estimates over transcript-level analyses in RNA-seq data analysis. The authors argue that gene-level estimates are more accurate and robust, providing better statistical performance and interpretability. They demonstrate that differential isoform usage can lead to inflated false discovery rates in differential gene expression (DGE) analyses on simple count matrices but that incorporating offsets derived from transcript-level abundance estimates can address this issue. The study also shows that the problem is relatively minor in several real data sets. To facilitate analysis, the authors provide an R package called *tximport*, which integrates transcript-level abundance estimates from common quantification pipelines into count-based statistical inference engines. The paper includes simulations and real data sets to support the claims and provides detailed descriptions of the methods used. Overall, the findings highlight the importance of carefully specifying the research question when selecting a statistical approach and suggest that gene-level analysis is often more suitable for detecting changes in overall gene expression.The paper "Differential analyses for RNA-seq: transcript-level estimates improve gene-level inferences" by Charlotte Soneson, Michael I. Love, and Mark D. Robinson discusses the advantages of using gene-level abundance estimates over transcript-level analyses in RNA-seq data analysis. The authors argue that gene-level estimates are more accurate and robust, providing better statistical performance and interpretability. They demonstrate that differential isoform usage can lead to inflated false discovery rates in differential gene expression (DGE) analyses on simple count matrices but that incorporating offsets derived from transcript-level abundance estimates can address this issue. The study also shows that the problem is relatively minor in several real data sets. To facilitate analysis, the authors provide an R package called *tximport*, which integrates transcript-level abundance estimates from common quantification pipelines into count-based statistical inference engines. The paper includes simulations and real data sets to support the claims and provides detailed descriptions of the methods used. Overall, the findings highlight the importance of carefully specifying the research question when selecting a statistical approach and suggest that gene-level analysis is often more suitable for detecting changes in overall gene expression.