What type of database does gnomAD use?

Question

[Mariana Ribeiro]

Hello!

I'm somewhat new to the field of bioinformatics, thus I needed to know, within the scope of my master's project, which database type gnomAD truly uses. I did a great amount of research, but I still don't really comprehend it. I am aware that the gnomad data is accessible via Google Cloud and can be explored using either Hail or BigQuery.

Note: To give some context, I have my own data, so I would like to try to create my own type of exome aggregation database, and the idea was to try to replicate gnomAD and take advantage of its open code source, eventually being able to load my data into it.

But I'm attempting to understand some details first.

Answer 1

[Jeremy Leipzig]

If you primarily want a variant warehouse that supports genomic region and sample queries at biobank scale, you should look into TileDB-VCF. TileDB-VCF offers Hail integrations to perform GWAS and other analyses. You would still need to fire up your own Spark cluster to support Hail. Perhaps more important for your application is that TileDB is a real open-source database that can work in a variety of cloud and cluster installations.

Answer 2

[Matthias Zepper]

I suppose there are multiple? Queries via Hail are very likely run against a Matrix Table, but the web application likely uses a different architecture (some graph database perhaps?). I would contact the Team at Broad for a definitive answer and some suggestions.

Answer 3

[Pierre Lindenbaum]

Google Cloud SDK https://github.com/broadinstitute/gnomad-browser/tree/main/deploy , but previous versions used mongodb https://github.com/macarthur-lab/gnomad_browser#mongo

[Pierre Lindenbaum]

I have my own data, so I would like to try to create my own type of exome aggregation database, and the idea was to try to replicate gnomAD and take advantage

IMHO, the gnomad browser could be complicated, if you just want to expose your VCF data.