I have biological replicates: they come from the same raw sample (tube of sediment) but from different DNA extraction & sequencing experiments. These are metagenomes and I used whole genome sequencing.
If I treat them separately, the abundances are consistent, some of them cluster together on a PCoA but not so well on hclust.
1) Should I concatenate the raw fastq files together beforehand?
2) Or should I treat them separately until after the alignment to a db step and then sum the counts of each replicates together?
What is the best procedure?