Is it possible to identify disease associated SNPs by comparing a single normal RNA-Seq data set with disease RNA-Seq data set ?
And are there any meaningful questions we can answer (except general differentially expressed genes and exon-junctions analysis)
Is it possible to link genotype and phenotype data ?
Thanx R
You're not giving us much detail to go on, but fundamentally with a single case and a single control you will have very little to work with no matter what kind of assay you have performed.
If by single you mean you have one case sample and one control sample, then the answer is no. Under a standard analysis, you would like to test the hypothesis that a given locus is associated with disease more frequently than you would expect to see by chance. Even if you assume your genotypes are perfect, that there is a SNP associated with your disease, and that this SNP is in a coding region, you will see many polymorphisms by chance between any two humans. You have no basis for asserting that any particular SNP is definitely associated with disease, even if that SNP is found only in your case, not in dbSNP, etc. You will expect to see even gene-truncating mutations even in normal humans (see the recent 1000 genomes papers for some quantification of this), so identifying a mutation that ablates a given gene's function only in your case would not be definitive (though it may suggest a candidate). This is the reason why genome-wide association studies require thousands of cases and controls for common, non-familial diseases.
As a special case, there are Mendelian diseases where possession of a single mutation in a coding region is causal for the disease with complete penetrance. Even in that case, you will still at best be generating a very large list of candidate mutations. In those cases the most powerful approach is usually to use family pedigrees to identify loci linked to disease (thus the name linkage studies). You will expect to see mutations introducing stop codons even in normal humans (see the recent 1000 genomes papers for some quantification of this), so identifying a mutation that ablates a given gene's function only in your case would not be informative with only one case.
Assuming the most straightforward reading of your third question, to get some answers on linking genotype and phenotype, the answer is yes, it can be done. Start with an introductory genetics textbook.
It is possible to discover disease associated SNPs even from a single disease sample, if the disease is known to be caused by a couple of rare mutations (e.g. some radiation diseases as I remember). I have seen one work in this line and the result looks meaningful to me. Nonetheless, for most other diseases, a single case-control pair is insufficient.
I'd be interested to see the paper. There will be a LOT of work coming out using sequencing approaches on rare highly penetrant diseases to find candidates (e.g. http://www.ncbi.nlm.nih.gov/pubmed/21572417, brand new NG paper on autism) but I think they will mostly use families, at least trios.
Yes it is possible, but you would be missing the SNPs from non expressed genes and skipped exons.
You can do the RNAseq at different time points of a disease process and do dynamic gene expression analysis to identify things like early driver genes, gene sets that share same expression profile and might behave similarly etc. RNAseq data coupled with microRNA data can be extremely useful for identifying the suppression/silencing of gene expression.
Here is a paper you should read: Genotype and SNP calling from next-generation sequencing data by R Nielsen, YS Song, et al (2011) Nat Reviews Genet 12:443. This is a review of next-generation sequencing methods used for SNP and genotype calling.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
The problem is that in biology a single sample is almost as good as none. So, I suggest the answer to all 1-3 is unfortunately, "in principle yes (given sufficient sample size), but with one sample: no".