Doesn't matter. Kraken can be and usually used on unassembled data and it taxonomically profile the sample based on Kmers. I would say these are diverging steps from the QC and cleaned data and are independent to each other.

It is up to you. Both analyses are valid. You can even run Kraken2 before and after assembly and compare the results.

Hi Valentina.

Usually, kraken2 or kaiju are used to do taxonomic identification of all the reads.

Assembly can be done independently of that and you can, again, used kraken or kaiju after to perform taxonomic identification of the assemblies.

Send me a pm if you need any help.

Thanks for your replies.

I have run the alignment with and without assembly, for the same sample, and compared the results.

For the assembled sample, after the alignment, I got less taxa hits compared to the not-assembled one.

I think that if you run an alignment after assembly, it will exclude many taxa because only the best represented ones can successfully go trough the process.

(I am talking for metagenomic samples here)

Valentina Vanghi that´s the same type of results I usually obtain from my metagenomic samples. After assembly, you are only considering the ones for which you can "isolate" the complete genome, and that are represented in higher degree. Before assembly, you are considering all the reads present in the sample, some of them correspoding to less-represented organisms (very low coverage), or organisms from which only a small region of the genome was sequenced.

Good luck with your analysis.

Hi Ricardo,

I am stuck with another thing now. Could you maybe help with it too?

https://www.researchgate.net/post/Kraken2_custom_db_How_to_deal_with_the_unmappedtxt_file

Thanks