1_all vs 1_lvl

[mars188]

Hello,

I merged several kraken2 reports using this command:

combine_kreports.py

It works fine but I am confused about let's say first sample has two columns i.e. 1_all and 1_lvl. I read it that 1_all is reads from Sample 1 rooted at this clade and 1_lvl is reads from Sample 1 at this clade. I am confused about the difference between the two as number of reads are also different for them.

Also, which column should be used for downstream analysis?