Unmapped reads - relative abundance and absolute counts in gene families output


I ran the following codes

1. to get relative abundance

–metaphlan-options “-t rel_ab --index mpa_vJan21_CHOCOPhlAnSGB_202103
–bowtie2db /home/smeeta.shrestha/metaphlan_data”
–input merged_S1_decon_mm10rm.fq.gz
–output merged_S1_decon_mm10rm_vJan21_RelAb_results

2. to get absolute counts

–metaphlan-options “-t rel_ab_w_read_stats --index mpa_vJan21_CHOCOPhlAnSGB_202103
–bowtie2db /home/smeeta.shrestha/metaphlan_data”
–input / merged_S1_decon_mm10rm.fq.gz
–output merged_S1_decon_mm10rm_vJan21_AbCount_results

However when i look at the genefamilies.tsv output unmapped reads

i get the following :

t rel_ab_w_read_stats|genefamilies|unmapped|3441838|
t rel_ab|genefamilies|unmapped|3436305|

ie there is less unmapped in relative abundance (3436305) compared to absolute abundance (3441838).

~ 5333 difference.

Any insights as what is happening.

Thank you

Can you compare the logs for the two HUMAnN runs for other differences? For example, if another species was selected for pangenome mapping from the rel_ab_w_read_stats run relative to the rel_ab run, that could explain why additional reads were mapped.