Predicting aberrant splicing

SpliceScan II tool running at BIG could be found here.

Aberrant splicing modeling plugin for Savant genome browser

Here is the link to the latest version of the plugin for Savant genome browser of v 2.0 and higher. The plugin contains all the source code, please feel free to modify it further for your needs.


The plugin is based on modeling of splice site junctions with the Bayesian splice site sensor that has been found to have an excellent sensitivity and specificity predicting splice sites compared to other sensor designs. Our plugin represents acceptors as red triangles and donors as blue or maroon colored (for GC non-canonical donors) triangles. The strength of a signal (how much signal triangle is filled with a color) characterizes its splicing competence. The stronger a signal, the more likely it flanks a true rather than cryptic exon. Here are the files containing some of the artificial decoy and real genetic variants in VCF format to test the plugin. Hypothetical changes disrupting ATM, NEB need to be correctly predicted by the plugin as splice site deleting variants. Also, the file contains verified real mutations disrupting splicing in BRCA1 and BRCA2 genes. An example of predicting changes associated with these variants could be found shown in the figures below. Please click on the figures to enlarge. Once you have questions please do not hesitate contacting me at . Here we model the effect on splicing of BRCA1 variants located in BRCA track disregarding common SNPs.

<Splicing> tab contains many general setups for the plugin.


<Variants> tab contains GUI controls to select variants we want to model from open VCF tracks.


In the following picture we observe several potentially aberrant splicing events associated with deletion of the splice sites or significant score changes of the splicing signal. Changes affecting cryptic splice sites could be seen by deselecting the <Report changes associated with annotated exons only> on the <Splicing> tab. You can click on the deletion, insertion or substitution button to relocate view the the variant. In order to freeze recalculation of the effects of the variants every time you relocate, you should deselect <Model effect of variants> on the <Splicing> tab.


When pressing a button corresponding to an affected splicing signal, that is located in the rightmost table column, plugin relocates to this variant.


The panel shown below displays modeling results with the SpliceScan II tool. Please note that SpliceScan II has been designed to model aberrant splicing with internal exons only. It is not suitable for the 3' and 5' most exons (first and last) modeling. SpliceScan II is capable of modeling not only the variants directly affecting splice sites, but also the effect of altering exonic and intronic enhancers/silencers that have been reported here. Please note that although method is sensitive, it is not very specific, which is the case with any other aberrant splicing prediction tools. All potentially deleterious variants need to be prioritized by score changes and verified in the wetlab, as could be exemplified by the collaborative ENIGMA study on aberrant splicing in BRCA gene.


When pressing a button corresponding to an affected exon, that is located in the rightmost table column, plugin relocates to this exon and it gets highlighted.


The plugin also displays the affected signal in GFF format.


These signals could be saved and later displayed in IGV genome browser.


Predictions in GFF format could be opened in IGV browser as shown below. The stronger the predicted exon compared to other overlapping exonic variants, the more chances this isoform would be present in mRNA.