Xenbase Image ID: 148694

Figure 2—figure supplement 2. Summary of RNA-seq data.(A) Correlation between RNA abundances from the replicate wild-type control samples, (B) correlation between RNA abundances from the replicate RFX2 knockdown samples, (C) MA-plot showing the relationship between fold-change and average abundance of each gene, and (D) Volcano plot showing the relationship of fold-change to FDR (adjusted p-value for differential expression). Genes differentially expressed in Rfx2 knockdown samples are indicated as red dots on (C) and (D). No systematic biases were evident among the differential expressed genes. Although the numbers of raw reads differed between samples, their normalized read counts correlated well. (E) Summary of sequencing data. The row titled ‘Total reads’ provides the numbers of Illumina Hi-Seq sequencing reads after pre-processing to remove low quality. We used the set of JGI 6.0 scaffolds longer than 10,000 bp for genome mapping, and the longest isoform of each gene model (‘Oktoberfest’ version), with bowtie1 (version 0.12.7) allowing two mismatches on the seed (-v 2 option). For mapping ChIP-Seq data to genomic scaffolds, we considered only unique hits (-m 1 option). For mapping RNA-seq reads to transcript models, we allowed for redundant hits (-an option) so as to maximize the signals from the RNA-seq datasets for the purposes of calculating differential gene expression, where redundant hits should not significantly affect the analysis, as each gene model was independently tested across conditions. We normalized across libraries by the total number of reads mapped onto a gene model. Subsequent tests of mapping without allowing redundant hits (-m 1 option) against the longest gene model confirmed that the differences between these two options was negligible. It should be noted that RNA-seq reads are paired-end 2 × 50 bp and ChIP-seq reads are single-end 1 × 35 bp.DOI: http://dx.doi.org/10.7554/eLife.01439.008 Larger Image Printer Friendly View

Return to previous page