Problems about Analyzing Joined Reads Using Qiime2

I am a new Qiime 2 user. When analyzing joined reads according to tutorial (, I met some troubles.

Firstly, the“manifest file”is a tab-seperated (i.e., .tsv) text file as the description in “Importing data” chapter ( However, in “Alternative methods of read-joining in QIIME 2”, it seems to be comma-separated.

When a .tsv text file was used, an error was reported.

When a .csv text file was used, data can be imported (although it cannot be viewed using qiime metadata tabulate).

Then, I denoised the sequences with Deblur. No error was reported. However, the results of feature table and rep-seqs seemed not right. Each feature was observed in one sample! It is unusual and I don’t know how to deal with it.

It would be great if some help can be provided.

Thank you in advance.

Welcome to the forum, @Jiaying!
The manifest you found in the read-joining tutorial predates the tab-separated SingleEndFastqManifestPhred33V2 format, and it looks like that format wasn’t compliant with our metadata format requirements.

Metadata tabulate is expecting a tsv, so it’s lumping all of your column headers together into one long string. That “super-header” doesn’t match the first-column header requirements, so it errors. The link in your error message describes those requirements. It should clear up why this is happening.

SingleEndFastqManifestPhred33 is dated, and should work properly for your data from “alternative methods for read joining”, but wasn’t built to handle .tsv. The newer SingleEndFastqMainfestPhred33V2 format is a .tsv and can be read as metadata.

I’m not totally sure why you’re getting that unusual behavior from Deblur, but it may be unrelated. Would you be willing to open a new topic about it, and include the complete sequence of commands you used to generate that result? I’d like to dig a little deeper and make sure there’s nothing misbehaving seriously.

I’ve opened issues on docs and qiime2 that might prevent confusion like this going forward.


  • Make sure you pass the correct file-format argument to the --input-format parameter.
  • Old (V1) SingleEndFastqManifestPhred33-formatted manifests can’t be read as metadata
  • please open a more-detailed question about how you got your deblur results in a separate topic

Thanks for letting us know about your situation!
Chris :flamingo:

Hi, Chris!

Thanks for your help!

I have corrected the file-format argument to the --input-format parameter. No error was reported, but the result of feature-table is inscrutably as previous (each feature was observed in one sample).

By the way, I have try to use DADA2 to apply quality control (as single end fastq), but the result was not changed. Therefore, I guess it was not caused by Deblur. Besides, although it was reported that QIIME2 was installed successfully and all data in the Tutorial (except section about joined reads), when I activate it, “complete:13: command not found: compdef”is reported.

I have opened a new topic about it named“Unusual feature table generated by joined reads without errors reported”(data and screenshots were attached).

Thanks again!



Hi @ChrisKeefe!
You are right! When using Deblur, “–p-left-trim-len 15” was missed. Now, all is OK!

And I think it may be helpful if this point can be mentioned in the Tutorial (this parameter is not mentioned in the joined reads section).

Thanks again!


1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.