As-is, a JBrowse track involves downloading chunks of the VCF file and parsing this on the client. Showing the individual variants generally does not require having any knowledge of the sample genotypes. A large VCF with 4400 genotypes explodes in size (basically 4400 additional columns), and contains a ton of data we dont care about. However, we do need more detail for the variant popup specifically.
It would be great if we could configure a track to read most data from a URL with a 'slim' version of the VCF, and only then the user clicks a specific variant would we query an alternate URL containing the full VCF. this alternate URL would be a very surgical loading of data, from just ~1 site.
This could really help the original VCF-backed table view too.