Virus- and Interferon Alpha-Induced Transcriptomes of Cells from the Microbat Myotis daubentonii

Hölzer, Martin; Schoen, Andreas; Wulle, Julia; Müller, Marcel A.; Drosten, Christian; Marz, Manja; Weber, Friedemann

Around 70% of the quality-checked and trimmed reads could be uniquely mapped to the reference genome of the related M. lucifugus (Table S2 ), using appropriate tools Kopylova et al., 2012; Schmieder and Edwards, 2011) . Based on pairwise comparisons of the mean values of the three replicates, protein-encoding genes that were induced or repressed by an absolute log2 fold change (FC) R 2 and with an adjusted p % 0.05 (DESeq2 (v1.16.1) [Love et al., 2014] ) were defined as significantly responding to either Clone 13 infection or IFN. To test whether biological replicates cluster together, we performed various principal component analyses (PCA) using our PCAGO web service (https://doi.org/10.1101/433078). In general, and visualized by the PCA based on the top 500 variant genes ( Figures 1B and S2A ), we found the biological replicates to cluster together. The PCA revealed that the transcriptomes of the 6-h post Clone 13 infection samples are closest to the mock samples and that the IFN-stimulated profiles of both time points are highly similar. The highest variance in the data (PC1 with 67,77%) can be explained by the large expression differences between the mock and Clone 13 6-h samples compared with the Clone 13 and IFN-stimulated 24-h samples. At 24 h, the Clone 13-infected cells show an expression pattern that is clearly distinguishable from all other conditions. One of the Clone 13 24-h replicates clustered somewhat apart from the other two replicates (see Figure 1B ), but the difference between these accounts for only 3.3% (PC2; Figure S2B ) of the whole variation in the gene expression data when directly compared with the mock 24-h samples. A 2D-and a 3D-PCA-movie, visualizing the overall stable clustering and how the principal components change by adding more and more less-variant genes to the transformation, can be found in the online supplement (Videos S1 and S2). CLARK classification and Krona visualization (Ondov et al., 2011) revealed 1% of the reads with high similarity to the RVFV reference genome in this specific sample, compared with 3%-4% in the other two replicates ( Figure S3 ). In addition to our reference-based analyses, we have created a comprehensive de novo transcriptome assembly for M. daubentonii by combining the output of various suitable assembly tools Grabherr et al., 2011; Liu et al., 2016; Xie et al., 2014) according to the results of Hö lzer and Marz (Hö lzer and Marz, 2019).