Different metaphlan2 output from same fasta and fastq sequences

Hi community (@franzosa, @fbeghini, @NSegata )!!!
I am trying to compare profile from .fasta and .fastq files.
I had two .fastq.gz files which I’ve unzipped into .fastq files. After that I have converted two .fastq files into .fasta files (by “seqtk”). Then I’ve concatenated two .fastq files and analysed them with metaphlan2.
After that, I also concatenated the two .fasta files and then analysed with metaphlan2. Number of reads are same in two concatenated files.
But I’m seeing differences in both bowtie2out.txt and profile.txt files (attachment) obtained from concatenated .fasta and .fastq files.
Can you please tell me why am i watching such differences in my result? Which result should I rely on?

Thanks and regards,

When converting FASTQ into FASTA, the mapping quality line is removed and so, some quality control internally done by MetaPhlAn is no longer executed and all the reads are then used.
I’d suggest to use directly the FASTQs and feed them to MetaPhlAn by comma separating the input.

1 Like