Figure 2

RNA-Seq mapping results. (A) A bird’s eye view of global RNA-Seq mapping on each strand using the SW time point as an example. Spikes are mostly non-coding RNAs (ncRNAs). (B) Example of a normalized RNA-Seq mapping. Two annotated tRNA-Asn are represented by the green and cyan boxes. The two DNA sequences of tRNA-Asn are identical, which leads to RNA-Seq reads being “ambiguously” mapped to two locations. Therefore, the expression value was calculated by dividing the amount of reads by two. (C) Scheme illustrating the gene expression quantification algorithm. RNA-Seq coverage (the red curve) and CV (blue curve) per nucleotide are plotted. The colored vertical bars border the dynamic programming segmentations based on the CV curve. In this particular case, CV is divided into 42 segments. We adopted a threshold of CV =1.0 (horizontal dotted blue line) to filter out segments with CV<1.0, allowing us to keep quality segments for gene expression quantification. (D) Distribution of gene expression values (using the SW cell as an example) follows a power-law distribution. The probability density p(e) ∽ e^- a, where e is gene expression, is best fitted with α=1.74 (according to Clauset et al’s algorithm [22]). This panel shows the cumulative distribution Pr(E>e) of gene expression along with the power law fit exponent -α+1=-0.74. Genes with expression ranging from 15x to about 1000x (indicated by the slant) fall into the power-law distribution, in good agreement with a previous report [23].