Interaction studies using SPSS--urgent help needed

ashi

New Member
#1
Hi All,

Not sure how many people out there know much about genetics and statistical analysis. I'm trying to submit my thesis and this person in my group (who is no longer contactable) did some work for me regarding interaction studies using linear regression on spss. NOw having a look at it im quite confused how they did it and i have to do the same work again for one more table to finalise, so im quite stuck.

Basically i have two lets say variants or mutations in this gene im studying, and i want to test if these two interact to affect a phenotype (trait you can measure ie..lets say BMI of a person)

So my two mutations are:
Mutation 1 three possibilities: CC, CG, GG
Mutation 2 three possibilities: GG, GA, AA

So i want to see in a group of 1200 individuals if these interact with each other to affect another phenotype ie..BMI, which i also have for all the individuals.

So the statitician she did this for me using linear regression, and came up with a Beta value and a p-value for each SNP alone and one for interaction for the studied phenotype. Not sure how she got this (something like the table below)

mutation 1: beta value pvalue
mutation 2: beta value pvalue
Interaction: beta value pvalue

Anyone that may know please feel free to leave a response to show me how i can do this on spss...if possible a step by step one as even though i knw how to use it statistics is still confusing for me:)

Thanks for your help.

ashi
 

JohnM

TS Contributor
#2
I think if you set up the data table in a spreadsheet-type layout in SPSS you should be able to run a linear regression on it. Each column will be a variable (Mutation1, Mutation2, BMI, etc.) and there should be a row for each data point or "case."

See the attached image for an example....

Include the mutation1 x mutation2 interaction in the model.