1.4.1版本的变化我们在土星1.4.0报告一个错误。(Bioconductor版本3.15)。土星1.3.1中引入的错误是无意中(从前者Bioconductor猛击)。注意,错误并不因此出现在任何老Bioconductor 1.0版本3.13和3.14(土星。1.1 x。x和1.2.x)。错误细节:想象一个有三个亚型和两个细胞类型的基因。我们的目标是评估细胞类型之间的差。所有亚型表达的细胞类型1的所有细胞。然而,没有一个亚型表达的细胞在细胞2型(即。,基因不表达的细胞类型2)。土星计算的log-odds一定从池中同种型的亚型在每个细胞类型,然后比较这些log-odds细胞类型之间的估计。 However, in this example, the log-odds of picking a certain isoform from the pool of isoforms in cell type 2 cannot be computed, as there is no data. Hence, the DTU test statistic should be NA. However, due to erroneous handling of NA estimates, which was inadvertently introduced in satuRn 1.3.1. while aiming to resolve github issue 16, the log-odds in cell type 1 will be compared to zero. Hence, (erroneous) results can be obtained for this contrast, even when there are no data in cell type 2. Note that in many cases such isoforms may not pass filtering and would not get evaluated altogether. However, when analyzing sprase scRNA-Seq datasets with a lenient filtering criterium, this problem will apply, and will result in mistakes in the inference. Changes in version 1.4.0 satuRn version for Bioconductor release 3.15 Changes in version 1.3.1 - Bug fix: allow for fit errors to be propagated as NA results (github issue 15 by @jgilis) - Bug fix: handle experimental designs with empty factor levels correctly (github issue 16 by @XueyiDong) - Bug fix: identify transcripts that are the only expressed transcript of a gene and set NA results (github issue 17 by @jgilis) - Bug fix: handle extreme z-scores in testDTU with diagplot2 option - Enhancement: plotDTU now allows for sparseMatrix input Changes in version 1.1.2 - Improved error handling Changes in version 1.1.1 - Initial Bioconductor release of satuRn Changes in version 0.99.0 NEW FEATURES - Added a NEWS.md file to track changes to the package. SIGNIFICANT USER-VISIBLE CHANGES - Your main changes to a function foo() or parameter param. BUG FIXES - Your bug fixes. See more details at //www.anjoumacpherson.com/developers/package-guidelines/#news.