I am trying to work on SNPs case-control studies. I am working on benign and malignant Melanoma samples and our goal is to differentiate benign and malignant by looking at SNPs. I just want to get SNPs list that are unique in malignant but not benign.I am using gPLINK but I am not sure whether I am using the correct analysis for my SNPs data. The samples are unrelated to each other. I will show the steps what I figured out so far is the way to do in gPLINK for my analysis.

1. make bed files
2. Quality control steps: MAF, GENO, MIND and HWE
3. Again make bed file after filtering
4. Association tests: Allelic association tests or logistic regression(not sure which one to use)
5. look into p-values, OR ratio and/ chisquare values​
6. Manhattan plot- for significant snps
7. LD plot( also not sure about this)

Questions:

1. What kind of association test I should do if I want to just look at phenotypic effects or just genotypic effects or both?
2. Since those samples are unrelated to each other, I think I don't have to do stratification test right?
3. In which case it is good to look at LD plot?

Really appreciate your help!