JETTA was demonstrated in high-throughput mRNA sequencing (RNA-Seq) data of human liver and muscle tissue samples. JETTA took pre-processed RNA-Seq data as inputs, calculated alternative splicing signals, and detected alternatively spliced exons. As results, JETTA detected 207 skipped exons. Details of the input files and steps of the analysis are available here.
The RNA-Seq data was obtained from Xu et al, PNAS, 2011. The sequencing reads were mapped over exon and junction regions according to the original paper, and the expression indices of genes, exons and junctions was calculated in RPKM (Wang et al, Nature, 2008). This data set provides replicates with which alternative splicing signals such as MIDAS can be calcualted.
|Filtering||Description||#. Genes||#. Exons|
|Junction supports||Supported by at least one juction satisfying the same MIDAS criterion||1,037||3,410|
|Average gene expression of each liver and muscle||> 1 RPKM|
|# of reads on exons in over-expressed||> 20|
|# of reads on exons in less-expressed||= 0||88||137|
137 exons of 88 genes were selected for skipping events by
Download the list of the skipped exons
Plotting examples: SLK
We compared the detected alternative splicing events in RNA-Seq with the alternatively spliced exons detected from the GG-H array data (Xu et al, PNAS, 2011) (Download). The overlap ratio was calculated as (# of overlapped detections with the GG-H results)/(# of total detections in RNA-Seq). The left figure shows the overlap ratios of detections with and without junction supports according to MIDAS p-values. The right figure shows the overlap ratios when the same number of top candidates were selected according to MIDAS p-values.
Note that with p-values < 0.001, JETTA detected 996 candidates, among which 52% were rediscovered by microarray experiments.
Data, results and R scripts
Back to top