Validation of predicted mRNA splicing mutations using high-throughput transcriptome data



Playing this video requires the latest flash player from Adobe.

Download link (right click and 'save-as') for playing in VLC or other f4v compatible player.


Recording Details

Collection/Series: 
PIRSA Number: 
14050059

Abstract

This work has been published:Viner C Dorman SN Shirley BC and Rogan PK (2014)Validation of predicted mRNA splicing mutations using high-throughput transcriptome data [v1; ref status: indexedhttp://f1000r.es/2no]F1000Research20143:8 (doi:10.12688/f1000research.3-8.v1)Additionally this work has been accepted for a highlights presentation at the upcoming Great Lakes Bioinformatics Conference (GLBIO) in Cincinnati Ohio and it was recently presented as a poster at London Health Research Day (LHRD).Abstract:Interpretation of variants present in complete genomes or exomes reveals numerous sequence changes only a fraction of which are likely to be pathogenic. Variants predicted to alter mRNA splicing in particular can be validated by manual inspection of transcriptome sequencing data however this approach is intractable for large datasets. We show that abnormal mRNA splicing patterns are characterized by reads demonstrating either exon skipping cryptic splice site use and high levels of intron inclusion or combinations of these properties. This paper presents Veridical an in silico method for the automatic validation of DNA sequencing variants that alter mRNA splicing. Veridical leverages large numbers of control samples (that lack a putative mutation) applying z-tests to Yeo-Johnson transformed data to normalize read counts of abnormal RNA species in mutant versus non-mutant tissues. With the transformed data the null hypothesis based mainly on either counts of intronic or junctional reads can be rejected for true splicing mutations using conventional parametric statistical methods.