Running multiple samples while loading index only once 

Hi,
I currently have a 200Gb database and it takes some time to load it. If this process could occur only once, it would save quite some time if I am to classify over 50 samples. 

Centrifuge has this feature, maybe some ideas can be drawn from there:  
https://github.com/DaehwanKimLab/centrifuge/pull/95