Reverse "normalize by number of reads in the sample"

Hi! How can we get counts not normalized by the number of reads in the sample, while still normalized by read length and sequence depth? Do I simply multiply the normalized counts by the total number of sequences in my metagenomics sample fasta file?