Hi
I have seen David Fast posted this same error, but i am not getting where i am wrong in my steps. could someone help me figure this out?
I am using qiime2-2019.10, with conda.
I am trying to merge the information of 2 runs. So i run dada2 on each of the runs separately, and then merged the tables and the representatives sequences using "qiime feature-table merge" and qiime feature-table merge-seqs". After, I run feature classifier with a pretrained silva classifier filtered for v3 -v4 region:
qiime feature-classifier classify-sklearn
--i-classifier classifier_silva_v3_v4.qza
--i-reads run1_run2/merged_merged_run1single_run2single/merged_rep-seqs.qza
--o-classification run1_run2/merged_merged_run1single_run2single/merged_taxonomy.qza
I was able to run qiime diversity core-metrics etc , but when i run "qiime taxa collapse" with the merged_representatives_sequences and the merged_taxonomy.qza it gives me the error posted:
Feature IDs found in the table are missing from the taxonomy.
If instead of running dada2 for each of the independent runs, I merge all the raw sequences of both runs and do the whole process ( dada2, taxonomic classification etc), i do not have this problem in the collapse step. However, I realized that core metrics analyses ( bray-curtis pcoa etc) gives me a very different result. It is my understanding that merging two runs before dada analysis might not be right, cause sequence quality algorithm of dada2 would be slightly different for both of the runs, is that correct? Then, this strategy would not be an option...
Welcome! Iām not an expert, so take this with a grain of salt, but two things come to mind. First, dada2 documentation advises āYou should only merge tables if the exact same parameters are used for each run.ā So Iād check that your dada2 parameters were identical to begin with.
Secondāand this could just be a typo in your questionābut you said you ran qiime taxa collapse with āmerged_representatives_sequences and the merged_taxonomy.qzaā. But taxa collapse needs a table and a taxonomy as inputāqzas of the types FeatureTable[Frequency] and FeatureData[Taxonomy]. What exactly did you type to get this error? Because when I ran taxa collapse with a representative sequences file instead of a feature table, my error was ā(1/1) Invalid value for āāi-tableā: Expected an artifact of at least type FeatureTable[Frequency]. An artifact of type FeatureData[Sequence] was provided.ā
Thank you so much for your response, and sorry for the late reply!
I had a look on my dada2 parameters and the length truncation was indeed different between runs. So i thought i found what the problem was, and started over the analysis specifying the same paremeters in both runs. Then I merged the tables and representative sequences, and assigned taxonomy to the merged files.
But, to my frustration, I am getting the same error Feature IDs found in the table are missing from the taxonomy: when i try to collapse the taxa using the "merged feature table" and the " taxonomy table generated after running "feature-classifier".
@wburgess's pointer may not have solved your issue, but it was sound advice. Once you resolve the error you're dealing with, be sure to stick with the same DADA2 parameters whenever you're merging after denoising.
This error message indicates that some of the Feature IDs in your table don't exist in the taxonomy you're using. This could happen if you used a taxonomy built with some but not all of your features (e.g. if you built the taxonomy before merging), if your taxonomy was built from a filtered set of sequences (I've done this!), and occasionally in other fun ways.
Please take a look at the provenance tab of your QIIME 2 artifacts in q2view and see if you can puzzle out what went wrong. If you can't figure it out, please share the artifacts you're using as inputs when you get this Error message, and we'll figure out it out together.
