How do I merge Framingham Heart Study phenotype data from BioLINCC with genotype data from dbGaP using participant IDs?

Hi everyone,

I'm trying to understand the workflow for combining Framingham Heart Study phenotype data from BioLINCC with Framingham genotype data from dbGaP.

Our lab currently has access to the phenotype data (sleep studies, MRI, cognition, dementia outcomes, etc.) through BioLINCC. We are considering applying for access to the corresponding genetic data in dbGaP so we can analyse AQP4 genotype together with these phenotypic variables.

Before we submit the application, I'm trying to understand exactly how the two datasets are linked.

My questions are:

After access to the dbGaP dataset is approved, how do you identify which genotype belongs to which participant in the BioLINCC dataset? Do BioLINCC and dbGaP use the same participant ID, or is there a separate mapping file that links the IDs? If there is a mapping file, what is it called and where is it obtained? Has anyone merged Framingham phenotype and genotype datasets before, and are there any common problems or things to watch out for? Is there any documentation or published workflow that explains this process?

I'm not asking for controlled-access data or code. I'm only trying to understand the standard workflow before we apply for dbGaP access.

Thanks!

biolincc framingham dbgap genotype phenotype

• 44 views • link

updated 5 days ago by SeqBench •

• written 1 week ago by Krishalls •

1 answer

The join key for FHS is shareid. dbGaP maps its own internal dbGaP_Subject_ID to the submitted subject ID, and for Framingham the submitted ID is the shareid, which is exactly what your BioLINCC phenotype files carry. So it's a shareid-to-shareid merge via dbGaP's own mapping files, nothing you have to construct.

The two files you want from the dbGaP download are the Subject file (dbGaP_Subject_ID to shareid, plus consent group) and the SubjectSampleMapping file. You need the second one because the genotype files are keyed on sample IDs rather than subject IDs.

Main thing to watch: expect a partial merge. BioLINCC and dbGaP are separate approvals with different consent subsets, so not every phenotyped participant will have genotypes available to you. Also check the generation split if any of your phenotype files are older releases, since Original/Offspring/Gen3 had separate ID spaces before shareid unified them.

• 0 views • link

written 5 days ago by SeqBench •