Kraken2 seems to classify reads from read pairs separately

[SimonHegele]

Tried to classify samples from paired-end sequencing using the command:

kraken2 --db {database} --paired {query} {mate} --report {report} --threads {threads} > {out} 2> {log}

My samples had 60-80 million read pairs, however Kraken2 reported 120-160 million fragments (two times as much).
Apparently each read is counted. At the same time same taxa have uneven numbers of associated fragments.
This indicates that the reads of a read pair could be mapped to different taxa.

[jenniferlu717]

I believe your command is incorrect. It should be

kraken2 --db {database} --report {report} --threads {threads} --paired {query} {mate} > {out} 2> {log}

Read files should be given last. and do not need to be right after the --paired argument