Developed and Maintained by Kaoru Ito (splicing.variant[at]gmail.com)

This document is for the paper "Identification of Pathogenic Mutations in LMNA and MYBPC3 That Alter RNA splicing" (Kaoru Ito and Parth Patel et al.) in PNAS 2017.
In the paper, we chose possibile splicing variants by in-silico prediction and tested them by cell-based splicing assay. Several perl and R scripts placed here were employed to perform the analysis.

*** In the paper, we focused on splcing variants that create / lose splice site. If you'd like to know about exon-skipping variants, please check SPANR (http://tools.genes.toronto.edu/)***

In the intepretation of genetic variants, synonymous mutation is considered benign. Also missense mutaiton is on the fence in terms of its deleteriousness, whereas stop-gain/loss, framn-shift and splice-site (GT/AG-broken) mutations are considered damaging. These intepretations are NOT ALWAYS TRUE.

Messenger RNA splicing, where intron regions flanked by a splice donor site and a splice acceptor site are cut out, occurs before protein syhthesis by mRNA translation.
In the mRNA splicing procedure, it's known that a nucleotide alteration that doens't change amino acide sequence can activate a cryptic splice site, which results in creating an aberrant splice donor/acceptor site in the middle of an exon / intron, followed by the exon truncation / intron retention in the protein. Additionally, not only GT/AG broken but also nearby nucleotide changes (donor:-3~+6bp, acceptor:-20~+3 from the splice juction) can disrupt the function of a canonical splice site.

Several papers reported that such variants cause severe mendelian disroders, such as progeria syndrome and dilated cardiomyopathy However, because a method for detecting such splicing-altering variants in a high-throughput pipeline had not been developed, these variants were overlooled in usual NGS settings.

Here we present a high-throughput-friendly method to detecte candidates for a splicing-altering variants. Additionally we developed a cell-based splicing assay utilizing NGS technology for mulitiplexing analysis, where construct design and assessment of splicing alteration are automated.

1. Calculate scores to assess the possibility for a splicing variant and choose candidates for cell-based splicing assay
   -- Regress_Score.v0.##.R
   
2. Design constructs for the candidates to perform cell-based splicing assay
   -- ConstructDesigner.v0.##.R
   
3. Perform cell-based splicing assay (transfect constructs to cells, RNA extraction from the cells, prepare libraries for Miseq run
   
4. Count normal / aberrant splciing products and calculate p-values from raw-fastq files (alignment not required)
   -- Make.inputframe.#.pl
   -- SpliceConstructSearchGrepV1.##.pl
   * # means developing version. Please use the latest one.

To run each script, please read README_regress.score.md, README_construct.designer.md and README_spliceconstructsearch.md.

Because p-value of the assay reflects the degree of aberrant splicing which changes continuously, we investigated the relationship between the p-values and clinical diagnosis in variants from clinical databases (Supplementary Fig.). As expected, we found a significant correlation between -log10 p-value of the assay and severity of the clinical diagnosis (missense variants included p=8.2e-06, missense variants excluded p=1.2e-06, kendall’s rank correlation test). Since clinically likely-benign and benign variants never surpassed p<0.001 and just missense variants (which can be deleterious without affectingh splicing) in likely-pathogenic and pathogenic groups had p>0.001, we defined the threshold p=0.001 to distinguish pathogenic variants from benign variants in terms of a splicing variant.