Uchime de novo filtering (chimera removal) error

Hello, I am following the Qiime2 tutorial “Identifying and filtering chimeric feature sequences with q2-vsearch”. I am doing this because even though I used dada2 on my data, when I went to upload my fasta file to NCBI, it flagged 1606 chimeras. When I used the commands below, I received an error message:
There was an issue with loading the file uchime-dn-out/chimeras.qza as metadata:
Metadata file path doesn’t exist, or the path points to something other than a file. Please check that the path exists, has read permissions, and points to a regular file (not a directory): uchime-dn-out/chimeras.qza

qiime feature-table filter-features
–i-table mesocosm-filtered-table.qza
–m-metadata-file uchime-dn-out/chimeras.qza
–o-filtered-table uchime-dn-out/mesocosm-filtered-table-nonchimeric-w-borderline.qza

qiime feature-table filter-seqs
–i-data rep-seqs-mesocosms.qza
–m-metadata-file uchime-dn-out/chimeras.qza
–o-filtered-data uchime-dn-out/rep-seqs-mesocosms-nonchimeric-w-borderline.qza

qiime feature-table summarize
–i-table uchime-dn-out/mesocosm-filtered-table-nonchimeric-w-borderline.qza
–o-visualization uchime-dn-out/mesocosm-filtered-table-nonchimeric-w-borderline.qzv

I have attached the chimeras.qza file. BTW, these are the exact same commands used in the tutorial with my own file names inserted.

Also, If you could also comment why I would have so many chimeras present after using the dada2 command? I did read some previous forum discussions on this but the only thing that I took away from that was that dada2 should be sufficient if you are applying it to ASV data instead of OTU data, which is what I did.


Hi @Lisa_Crummett,
This is an indirect answer to your issue.

You really should be uploading raw .fastq files not processed fasta files. Who knows how well the current techniques will hold in the future, it’s always possible/desriable to re-analyze data from raw reads rather than some processed ones.
As for why something is considered a chimera, and whether or not it is correct or not, this is a very complex discussion with no right answer. If we truly knew what was a chimera or how to correctly distinguish them from a real read you can bet that there would be no differences in methods and everyone would simply just remove them the same. I find DADA2’s default chimera detection to be actually pretty good, even on the conservative side of things so I’m not sure what is raising those errors in NCBI. Maybe it’s flagging something else as a result of your processing?
Anyways, while some comments more directly regarding your error, I would recommend considering not uploading fasta files.