Hi,
I'm a Computer Science postgraduate new to the area of bioinformatics. I have been given a dataset of a couple of SNPs with 600 subjects. I've a quantitive value, measuring disease severity.
What tools should I use to see if there is a correlation?
I'd really appreciate your suggestions.
So you can ideally compare the number of people with each allele and also tabulate stats on the distribution of disease scores among each allele, stating which allele you conclude is the disease causing allele and what kind of mutation (transition/transversion) it is.
Also, to spice stuff up, maybe add some ANOVA into the mix? :)
I'm moving this to an answer, an ANOVA (or linear model more generally) is the simplest way to handle data like this.
@davidswordster: Keep in mind that we're assuming the 0-200 severity scale units represents relatively evenly spaced increases in severity. While this is very likely the case, you might want to double check with the people who generated the data to ensure this (if the scale were more like 0-7 then you'd need to use ordered logistic regression instead).
Thank you so much for your answers. I'm looking into these as we speak.
Just one other thing, and it was something that was mentioned to me, but not elaborated to me before, I've studied Hardy-Weinberg and it seems pretty straightforward, but something has confused me about the suggestion. I was asked to generate p-values for each SNP, which is no problem, but the way that it was suggested to me denoted that there would be some indication of interesting SNPs based off of this. Is there a way to determine SNPs of interest based on the dominant allele frequency?
I'm not sure how that would work given how your dataset is structured. In general, a population with a mendelian disorder caused by a given variant should show a violation of Hardy-Weinberg at that variant position. Presumably that's what was being implied, however applying that principle to a highly graded disorder is probably not a very fruitful avenue to pursue.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Ummm, what do you mean when you say you have quantitative value? And, what data do you have on the SNPs exactly? The alleles involved? The frequencies they are seen in?
Thanks for getting back to me.
Let's say I have a value between 0 and 200 that represents the severity of the disease, and I have this value for each of the 600 subjects.
Then for the SNPs, I've the alleles for each subject.
Alright, is the reference allele given to you or is it just this data (Subject, Allele, Disease-Score) from which you're supposed to infer that?
Can you describe what you mean, by reference allele?
It's the allele found in the reference genome at the exact same position as your Single Nucleotide Variant.
Yes, I believe so.