Hi! Guys, I want to do some association researches between SNP genotypes and phenotypes in cancer patients, just like this article for investigating association between SNPs in UGT2B and breast cancer.

I know SNP information is located in somatic mutation, while I have downloaded somatic mutation MAF files, there are 5 subset files. I leave them as here.

1. BCGSC__IlluminaHiSeq_DNASeq_automated
2. BCM__IlluminaGA_DNASeq_automated
3. BCM__Mixed_DNASeq_curated
4. BI__IlluminaGA_DNASeq_automated
5. UCSC__IlluminaGA_DNASeq_automated

Actually these files are all somatic mutation files, but sequenced by different institutions and platforms. There are also some differences between these files, for example, here is a part of BCGSC__IlluminaHiSeq_DNASeq_automated file (11,12 13column), the genotypes of SNPs in Tumor_Seq_Allele are all zygosity.

Reference_Allele    Tumor_Seq_Allele1    Tumor_Seq_Allele2
G    A    A
G    A    A
A    G    G
A    T    T
C    A    A
C    T    T
A    G    G
..

But in BI__IlluminaGA_DNASeq_automated file, the mutations are all heterozygous.

Reference_Allele    Tumor_Seq_Allele1    Tumor_Seq_Allele2
G    G    A
G    G    A
A    A    T
C    C    T
A    A    G
A    A    G
G    G    A
..

Although having read the Tutorial Working with MAF files from the TCGA, I still have no idea which file to choose.

To conclude, I have two major problems troubled me.

Firstly, can Tumor_Seq_Allele 1 and Tumor_Seq_Allele 2 really represent the SNP's genotype? Because in the next step , I will select TagSNPs based on MAF(Minor Allele Frequency, filter criterion >0.05), but if I acquire genotypes like that, all the SNPs' frequency are below 0.05, which means no TagSNPs! I'm not sure whether it is right, this question also mentioned the Tumor_Seq_Allele, but I don't understand REF/ALT allele and how to get a more reliable SNP genotype. Secondly, the great disparities between BCGSC and BI files make me confused which file to choose for my next step.

I hope you guys can give me some suggestions. Many thanks!