Division of Bioinformatics and Biostatistics, National Center for Toxicological Research, United States Food and Drug Administration, Jefferson, Arkansas 72079, United States.
Chem Res Toxicol. 2021 Feb 15;34(2):529-540. doi: 10.1021/acs.chemrestox.0c00368. Epub 2020 Dec 23.
While RNA-sequencing (RNA-seq) has emerged as a standard approach in toxicogenomics, its full potential in gaining underlying toxicological mechanisms is still not clear when only three biological replicates are used. This "three-sample" study design is common in toxicological research, particularly in animal studies during preclinical drug development. Sequencing depth (the total number of reads in an experiment) and library preparation are critical to the resolution and integrity of RNA-seq data and biological interpretation. We used aflatoxin b1 (AFB1), a model toxicant, to investigate the effect of sequencing depth and library preparation in RNA-seq on toxicological interpretation in the "three-sample" scenario. We also compared different gene profiling platforms (RNA-seq, TempO-seq, microarray, and qPCR) using identical liver samples. Well-established mechanisms of AFB1 toxicity served as ground truth for our comparative analyses. We found that a minimum of 20 million reads was sufficient to elicit key toxicity functions and pathways underlying AFB1-induced liver toxicity using three replicates and that identification of differentially expressed genes was positively associated with sequencing depth to a certain extent. Further, our results showed that RNA-seq revealed toxicological insights from pathway enrichment with overall higher statistical power and overlap ratio, compared with TempO-seq and microarray. Moreover, library preparation using the same methods was important to reproducing the toxicological interpretation.
虽然 RNA 测序 (RNA-seq) 已成为毒理学基因组学的标准方法，但当仅使用三个生物学重复时，其在获得潜在毒理学机制方面的全部潜力尚不清楚。这种“三样本”研究设计在毒理学研究中很常见，特别是在临床前药物开发期间的动物研究中。测序深度（实验中读取的总数）和文库制备对 RNA-seq 数据的分辨率和完整性以及生物学解释至关重要。我们使用黄曲霉毒素 B1 (AFB1)，一种模型毒物，研究了在“三样本”情况下，测序深度和文库制备对 RNA-seq 毒理学解释的影响。我们还使用相同的肝样本来比较不同的基因分析平台（RNA-seq、TempO-seq、微阵列和 qPCR）。黄曲霉毒素 B1 毒性的既定机制是我们比较分析的基准。我们发现，使用三个重复，至少需要 2000 万条读数才能引发 AFB1 诱导的肝毒性的关键毒性功能和途径，并且差异表达基因的鉴定在一定程度上与测序深度呈正相关。此外，我们的结果表明，与 TempO-seq 和微阵列相比，RNA-seq 从通路富集中揭示了毒理学见解，具有更高的统计能力和重叠率。此外，使用相同方法进行文库制备对于复制毒理学解释很重要。
