JETTA: Alternative splicing detection with human muscle and liver samples on the GG-H arrays


JETTA was demonstrated on the GG-H arrays (Xu et al, PNAS, 2011) with human liver and muscle tissue samples. JETTA processed raw probe intensities, calculated gene/exon/junction expressions, and detected alternative splicing events between two tissues. For skipped exons, JETTA found 145 confident candidates, among which three were seleceted for verification and validated by RT-PCR.

Microarray analysis and detection of alternative splicing events

Raw array data of liver and muscle samples was processed through GCBIN background modeling, median-scaling normalization and median-polish summarization. Alternative spcling events were detected by combined filtering of gene expression, DABG, MADS and MIDAS.

Filtering Description #. Genes #. Exons
Annotated alternative exons15,873165,586
MIDAS <0.01
DABG when over-expressed <0.01 5,53913,150
Junction supports Supported by at least one juction satisfying the same MIDAS and DABG criteria 2,9996,461
Average gene expression of each liver and muscle > median of overall expression
gene expression fold change <2
MADS <0.01
DABG when less expressed >0.1 122145

Detected candidates of alternatively skipped exons

145 exons of 122 genes were selected for skipping events by

Download the list of the skipped exons
Plotting examples: SLK, GARNL1, and VPS39

Validations with RT-PCR

Three detected genes, GARNL1(Exon25, PsId:368373), VPS39(Exon23, PsId:593948) and SLK(Exon13, PsId:38999), were validated with RT-PCR.

The left figure shows normalized fold changes( log2(muscle/liver) ) of array and RT-PCR (black: exon, read: inclusion jucntion, green: exclusion junction). All three candidates also have large fold-changes in the PCR results. The gel experiment results verified it again (right figure).


R script and outputs

Back to top