JETTA: Description

Junction and Exon Toolkit for Transcriptome Analysis (JETTA) is for Gene/Transcript Clust/Exon/PSR level expression index calculation and different alternative splicing detection. It is implemented using open source libraries and modified codes of

Supported Arrays

JETTA supports commerical Affymetrix exon arrays and exon-junction arrays following Affymetrix exon array design scheme.

JETTA also supports RNA-Seq data for alternative splicing analyses.

Expression Index Caculation

Expression indexes are calculated as the order of 1) Background Correction, 2) Normalization and 3) Summarization. In the Summarization step, background corrected and normalized probe intensities of a meta probeset are summarized to expression of the meta probeset. Meta probesets can be defined as gene/transcript clust/exon level.

1) Background Correction

JETTA estimates background signal using background probes and subtracts it from the probe intensity. Estimation of background signal is based on several models.

If the probe intensity is less than the estimated background signal, JETTA handles it according to options.

2) Normalization

Normalization of JETTA is done for core probes defined in probeset annotation file. If the PSA file is not specified, it considers all probes in the MPS files as core probes.

3) Summarization

Alternative Splicing Detection

JETTA detects alternative splicing events by providing DABG, MADS and MIDAS p-values and other useful information.

1) DABG: Detection Above BackGround

A DABG p-value indicates the probability of presence of a probeset(exon/junction). DABG p-value of each probe of a probeset is calculated as the ratio of background probes with higher intensities among all background probes with the same GC contents as the probe. Probe-level DABG p-value is summarized to probeset-level DABG p-value by Fisher's method. Sample group DABG p-value is given as geometric mean of DABG p-value of each sample.



MADS calculates significance of alternative splicing probesets between two sample groups based on background corrected and normalized probe intensities. It has several criteria to filter out transcript clusts and probes from the analysis.

Computation Issues

JETTA is designed for 32-bit machine with 2G memory, and it works best on MAC Pro 64bit with more than 4G memory. For TC expression calculation using MAT background correction, Median scaling normalization and LiWong model summarization with 9 Glue Grant arrays, it takes

Back to top