Entering edit mode
6.6 years ago
kelly.wang135
▴
70
Hi, I am doing allelic imbalance analysis (or allele-specific expression analysis) and have a question.
I want to see the association between marker SNP and nearby exon SNPs.
My data shows that for all three types of status of marker SNP (AA, AB, BB), AI ratios of exon SNPs are all nearly 1.
So, almost all exon SNPs should be removed as usual criteria for exclusion is 0.8 (SNPs with extreme major allele fraction above 0.8 are excluded).
So my question is, how can I conclude from this data? Does it show there is no association between the marker SNP and expression of nearby genes?
Thank you!
Hi!
Some questions, just to be sure I got it.
By exon SNP, you mean SNPs that are in the exons and that you genotyped via RNAseq?
"Ratio of exon SNPs are nearly 1" means that the MAF of the SNP is close to 0 (i.e. they are homozygous)?
Yes to both questions. Exonic SNPs are used to represent expression of that gene and expression can be assessed through RNA-seq.
AI is calculated as max (ref, alt) / ref+alt where ref and alt mean the number of ref or alt alleles. So AI=1 means homozygous alleles.
Thanks!
Last question (I think): for those homozygous SNPs, what is the genotype in DNA?
I ask because:
IF the SNPs are heterozygous in DNA and homozygous in RNA, then you have strong allelic imbalance (unless there's some kind of sequencing bias), i.e. what you want to see
If the SNPs are homozygous also in DNA then they are not informative.
Are you asking for thr marker SNP? Marker SNPs are AA, AB or BB, and exonic SNPs are almost all homozygous regardless of marker SNP.
Mmm... let's see: There is a good thread here on allele-specific expression. Maybe you could give a look to that: What is allele-specific binding, allele-specific expression, and allelic imbalance? When doing AI analysis you should start from SNPs for which your sample DNA is heterozygous.
Thank you. If there are only homozygous SNPs for the exons of this gene, I should conclude 1) ase analysis cannot be done with this data, or 2) there is no association between this marker SNP and the expression of gene?
I guess before attempting this kind of analysis, you should go through this video where it is very nicely and clearly explained the 'what' and 'why' aspects of allele specific expression (ASE)- http://medias01-web.embl.de/Mediasite/Play/7a15ffa352f64224a9cdd990c8bc64e01d
Furthermore, as Fabio mentioned, performing ASE studies on homozygous variant positions is logically wrong because if your variant caller has declared your site as homozygous (taking into consideration sequencing and mapping errors), your first step in ASE analysis should be to filter those sites and consider only biallelic heterozygous positions because technically (again taking errors into consideration), there is no second allele in your homozygous variant position for calculating ASE..
There is another very good paper on ASE commenting on some good practices for ASE analysis - https://genomebiology.biomedcentral.com/articles/10.1186/s13059-015-0762-6
Although, the above paper is based on RNAseq, some of the concepts mentioned there can be applied to DNAseq as well..