Metaphlan4 estimated_number_of_reads

Microbial community profiling MetaPhlAn
1

I found there is a discrepancy in estimated_number_of_reads between inputting two separate fastq files and inputting a single concatenated fastq file for paired-end data when using Metaphlan4.
I used cat command to concatenate paired-end reads.

Output for two separate fastq input:
the profile output for input of two separate fastq:
#mpa_vOct22_CHOCOPhlAnSGB_202212
#/software/MetaPhlAn/4.0.6/bin/metaphlan ./Clean/SRR14092160_1_Clean.fastq.gz,./Clean/SRR14092160_2_Clean.fastq.gz --input_type fastq --index mpa_vOct22_CHOCOPhlAnSGB_202212 --bowtie2db /resources/fdb/Software_bundle/MetaPhlAn4/database/mpa_vOct22_CHOCOPhlAnSGB_202212 --bowtie2out ./metaphlan4/SRR14092160.bowtie2.bz2 --nproc 12 -t rel_ab_w_read_stats --unclassified_estimation
#71940062 reads processed
#SampleID Metaphlan_Analysis
#estimated_reads_mapped_to_known_clades:59117153
#clade_name clade_taxid relative_abundance coverage estimated_number_of_reads_from_the_clade
UNCLASSIFIED -1 17.1838 - 12822909
k__Bacteria 2 82.81279 20.98874 59106513
k__Eukaryota 2759 0.00341 0.00086 10640

the profile output for input of single fastq:
#mpa_vOct22_CHOCOPhlAnSGB_202212
#/software/MetaPhlAn/4.0.6/bin/metaphlan ./Clean/SRR14092160_merge_Clean.fastq --input_type fastq --index mpa_vOct22_CHOCOPhlAnSGB_202212 --bowtie2db /resources/fdb/Software_bundle/MetaPhlAn4/database/mpa_vOct22_CHOCOPhlAnSGB_202212 --bowtie2out ./metaphlan4/SRR14092160.bowtie2.bz2 --nproc 12 -t rel_ab_w_read_stats --unclassified_estimation
#71940062 reads processed
#SampleID Metaphlan_Analysis
#estimated_reads_mapped_to_known_clades:59117218
#clade_name clade_taxid relative_abundance coverage estimated_number_of_reads_from_the_clade
UNCLASSIFIED -1 17.18374 - 12822844
k__Bacteria 2 82.81284 20.98876 59106578
k__Eukaryota 2759 0.00341 0.00086 10640

What causes the discrepancy ?

Thanks a lot!

2

This is interesting, could you trace back the inconsistencies to the t__SGB level?