1
$\begingroup$

I am writing a paper for non coding SNPs on patients with metastatic breast cancer. Having used a specific gene panel (NGS) of approximately 60 genes, I'm currently running out of ideas on what to write in the discussion section.

I am looking for suggestions based on somewhat efficient methods like a bioinformatics approach or specific textbooks revolving around RNA biology/biochemistry, in order to predict the outcome of non-coding SNP variants.

Some details:

  • The majority of the variants are intronic (~2300/3300), so I will mostly focus on splicing.

  • There are two commonly repeated SNPs in 3'-UTRs in over half of the patients, involving KRAS and CDKN2A.

Improve this question
$\endgroup$

2 Answers 2

Reset to default
1
$\begingroup$

A) Generate exon & transcript counts for your samples, library size + quantile normalize the data.
B) Group samples into variant/wild-type categories C) Use wilcoxon's rank sum test to see if there are differences in exon inclusion or overall expression. Remember that 3'UTR variants can also imply differential degradation by miRNA. Tools like TargetScan can help you with miR-analysis.

Improve this answer
$\endgroup$
1
  • $\begingroup$ You might be better off using something like DEXSeq to detect differential transcript usage $\endgroup$
    – fanli
    Commented May 17, 2016 at 0:00
1
$\begingroup$

You could try DeepSEA. It uses deep learning approach to predict function of noncoding SNVs. They use ENCODE and Roadmap Epigenomics for chromatin structure learning, 1KG for nonfunctional SNVs, HGMD for noncoding regulatory mutations, GRASP (Genome-Wide Repository of Associations between SNPs and Phenotypes) for noncoding eQTL and US National Human Genome Research Institute's GWAS Catalog for noncoding trait-associated SNVs. 'Predicting effects of noncoding variants with deep learning–based sequence model' paper describes the whole pipeline.

Improve this answer
$\endgroup$
1
  • $\begingroup$ I've checked this out before, but I'm mostly looking for a way to predict the length and probabilities (sequence too, probably) of individual transcripts stemming from the emergence of the alternative splicing sites (still based in machine-learning). Do you have any ideas ? $\endgroup$
    – civy
    Commented Mar 15, 2016 at 22:16

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .