I have a project where there is a list of genome-wide significant SNPs and we are interested in performing fine-mapping using only the summary stats data of GWAS. I was suggested to do a conditional analysis on single/multiple SNPs and also gene-wide analysis. I came across tools such as GCTA and VEGAS2 that accepts summary stats data.
After I run the analysis, I get a different set of pvalues and I plot the data from both the original and the new p values, the problem is I do not understand how to interpret the results. For example, in this paper http://www.thelancet.com/journals/laneur/article/PIIS147444221270234X/images?imageId=gr3§ionType=red&hasDownloadImagesLink=true , how can we interpret the results ? Do we think that the SNPs associated with Genes have independent association ? What happens if some SNPs that are in low LD (r2=0.2) appear to be significant (say 0.0005) after conditional analysis?
Also, for gene-wide analysis, VEGAS2 is the software that I came across. VEGAS calculates this gene P-value by aggregating the individual SNP P-values in that gene (using 1000G) while correcting for gene-length and LD. For example, the gene-wide analysis for HDAC9 (with 1000SNPs) has the following p value 6.23E-01 and PITX2 (28SNPs) has a p value 7.32E-01. Does this mean that the gene PITX2 is more significant than HDAC9 considering the number of SNPs?
and more importantly, why do we do such analysis and what do we intend to interpret from it?