A non-biologist asked me why this field doesn't use relational db's more. They have built-in indexing and nice query support and are widely used, etc. In addition, a database like postgres allows custom data-types and indexes so interval queries could be optimized.
My only thoughts were related to the number of records--there could be billions for even a single-sample BAM. Once you get to hundreds of samples, this becomes a lot of records. I don't know the practical limit for RDB.
What are the other reasons?
Why couldn't we be using relational database instead of BAM?