Hello, I have ran taxprofiler with success with th...
# taxprofiler
r
Hello, I have ran taxprofiler with success with the Metaphlan database. However taxprofiler provides only relative abundance counts, there is no parameter to estimate the number of reads coming from each clade (-t rel_ab_w_read_stats), is there any way I can use this parameter with nf-core/taxprofiler or I need to rerun the analysis ? Thanks
j
Yes! You need to add it to your `db_params`column in your database.csv
👍 1
r
Thank you !
a
Hello @James Fellows Yates and other interested on @Ruben Ladeira’s question! If you run it with -t rel_ab_w_read_stats in the db_params column in databases.csv then it outputs profiles with 5 columns: #clade_name clade_taxid relative_abundance coverage estimated_number_of_reads_from_the_clade and merge.py seems to be expecting 4 columns (see the error attachment). Is there any workaround to get anyway the full abundance table? The best would be that one could get the relative abundances table and the read counts table (for now we have our own scripts to join the abundances or counts), but it is that something that could be added to taxprofiler or is it more some issue/comment to do to Metaphlan developers? Perhaps you have at some point talked about that 🙂 Best, Albert , tagging others interested: @Juliana Assis @André Ferreira Cunha
❤️ 1
j
OK this one would be for @moritz!
The issue is the taxpasta rahter than metaphlan
I think we need to separate the two functionality (taxpasta) with the native table aggregations...
But you may have the native metaphlan table in your
results
table already?
a
@James Fellows Yates just fresh info, I see that despite the error message I got the combined table for metaphlan (metaphlan_db1_combined_reports.txt)!
🎉 1
So basically, yes to you question, I get the relative abundances that I want, but the run ends unsuccessful. What it would be great is to also be able to get the counts table, but not a big deal as we have an R script for that. Perhaps these post processing simple scripts could be just added as a way to combine the results without being part of taxpasta or triggering taxpasta?
j
Hrmm... So taxpasra processes whatever comes out of metaphlan, i don't think it coverts rel ab back to counts (iirc @moritz?) But if I'm understanding your request correctly - i seem to remember metaphlan could only produce one table at s time (counts OR rel ab)? Unless i am remembering badly or it's changed since then)?
So you would have to run metaphlan twice
Note I'm not against the idea of creating both both counts and rel abs , I'm just not sure technically how it work currently
a
Running taxprofiler with -t rel_ab_w_read_stats gives the estimated counts (estimated_number_of_reads_from_the_clade - see the attachment). and if I do not forget to turn on run_profile_standardisation (sometimes it happens 🙃) then I get the combined metaphlan relative abundances. It would be nice to also get the combined estimated counts. Just a small little improvement, but not a big deal as I said as we have our own script to extract the estimated counts from the metaphlan profile files.
✅ 1
j
@Albert Palleja Caro OK cool, (sorry for delay, I'm solo parenting this week as my partner is at a conferece)
Could you make an issue and I can look into this on Thursday (our normal dev hours : ) )
✅ 1
I think overall it makes sense 👍
m
I've added my thinking. We should always include the counts for MetaPhlAn. I'm curious to get your perspectives, though. https://github.com/nf-core/taxprofiler/issues/638#issuecomment-3227976559
👍 1
a
I replied there as well! Thanks for your comments