Error executing process > 'RelativeAbundanceReducedTaxa (1)'

[nbargues]

Hi,

I have an error during : Error executing process > 'RelativeAbundanceReducedTaxa (1)'

Here is the post I send to the qiime forum :

https://forum.qiime2.org/t/error-during-relative-abundance-step/16774

Thanks

[d4straub]

Hi there, QIIME2 forums are not the right place to discuss this, this is a specific issue with your data and ampliseq. To make it short: you seem to have no taxonomic classifications that go below level 2 (kingdom, I believe). Therefore, the ASV table cannot be used for taxonomic level 3 or above. This is not a problem with the pipeline but with your data, it seems. To make the pipeline finish gracefully, please add to your command --skip_ancom --skip_abundance_tables -resume. This should skip all steps where taxa are collapsed. Next, please make sure that everything is alright with taxonomic classification, it looks odd that your classification is so shallow.

[nbargues]

I know that is probably a classifier problem but I need the abundance tables, it's the reason why I'm using ampliseq. Do you notice something wrong with my command for creating my classifier ?

[d4straub]

No, these classifier commands look fine to me.

But your amplified region is quite long, more than 500bp. This might be tricky with MiSeq data. Do you have a sufficient number/fraction of reads passing DADA2? Check this in results/abundance_table/unfiltered/dada_stats.tsv: The numbers in the last column should have 50-90% of the numbers of the second column.

Also please check results/taxonomy/taxonomy.tsv, there you find your taxonomic classifications. Make sure these look fine. I doubt that this is the case.

[nbargues]

I performed a 2 x 300 read length on V1/V3 with 560 insert size and 30 overlap.

Yes you are correct, I have less than 5% in dada_stats.tsv and my taxonomy only have 0_Bacteria ....

Do you have an idea how to change that ? Thanks

[d4straub]

To increase that numbers you can basically use two strategies for DADA2: (1) truncate reads more aggressively, i.e. lower numbers for --trunclenf & --trunclenr. Problem is, you have not a lot of choice here. Still, look at result/demux/index.html and choose the lowest --trunclenf & --trunclenr that lead to 20bp overlap. (2) Allow more reads to pass the DADA2 quality filter (increase -ee). However, this feature isn't exposed in ampliseq. Another choice would be to forfeit DADA2 and use Deblur. This is also not supported by this pipeline. My plans are to expose DADA2's -ee so that there is one more possibility to tackle that problem. But I do not have a timeline here.

[nbargues]

Ok I run with what you recommend in (1). See if that change something. I've plan a new run of sequencing on Monday, but if I have the same result as this run, it seems pointless. What are your thought about what are wrong with my data ( ie quality , length , pcr duplicate ) ? I attach the multiqc report

multiqc_report.zip

[d4straub]

Quality scores seem too low. Test some more truncation values when you see that the choice that you made now improves results. i.e. reduce both by 5 and see if the overlap is still sufficient. If that isnt sufficient, you need to use other methods.

[nbargues]

My read length is 300 bp. After trimming with ampliseq, we see that the mean read length are 280 bp. So I run ampliseq with --trunclenf 150 and --trunclenr 150. With that I still have an error with this step but at the level 7 :

Command error: Plugin error from taxa:

Requested level of 7 is larger than the maximum level available in taxonomy data (6).

But I think that 150 is the limit for read of 280 bp and for having an overlap of 20. Do you think that I should truncate at 100 for example ?

[d4straub]

Indeed, after trimming with ampliseq, the mean read length is reduced by the primer length (which is removed, because non-biological data) and 300bp raw reads should be ~280bp.

Your previous information with (1)

I performed a 2 x 300 read length on V1/V3 with 560 insert size and 30 overlap.

and than this sentence (2)

So I run ampliseq with --trunclenf 150 and --trunclenr 150

do not fit.

If (1) is true, you really have an expected insert size of 560bp (including primer sequences! -> 560-20-20=520 excluding primer sequences), than --trunclenf 150 and --trunclenr 150 will only allow 280bp amplicons to be merged (150+150-20), which is <520bp. Therefore close to no sequences should pass the DADA2 step, because merging fails. The solution here would be to use something like --trunclenf 275 --trunclenr 265 = 540bp (520 amplicon + 20 overlap).

edit: If this is not what you mean, please share (a) the exact and complete command you use to start ampliseq and (b) the file results/abundance_table/unfiltered/dada_stats.tsv

[nbargues]

The new value solved the issue. Thanks !