Adding new taxa to a Kraken2 db

Question

[valentinavan]

Hi, can someone please check if these following steps are correct?

I am trying to add to my plants kraken2 db ("plant_original") few taxa genomes that I have downloaded from the NCBI website (alnus_glutinosa_GCA_003254965.1.fna, carpinus_fangiana_GCA_006937295.1.fna etc..).

for file in *.fna   
    do
        kraken2-build --add-to-library $file --db PATH/kraken/plant_original
    done    

Masking low-complexity regions of new file... done. Added "alnus_glutinosa_GCA_003254965.1.fna" to library (PATH/kraken/plant_original)

Masking low-complexity regions of new file... done.

Added "carpinus_fangiana_GCA_006937295.1.fna" to library (PATH/kraken/plant_original)

Masking low-complexity regions of new file... done.

kraken2-build --build --db ~/kraken/plant_original

Creating sequence ID to taxonomy ID map (step 1)...

Sequence ID to taxonomy ID map already present, skipping map creation.

Estimating required capacity (step 2)...

Estimated hash table requirement: 73390180936 bytes Capacity estimation complete. [1h53m11.000s]

Building database files (step 3)... Hash table already present, skipping database file build. Database construction complete. [Total: 1h53m11.000s]

Then, I run kraken with one of my sample against this new updated plants db (I did not change the name of the db, is still called plant_original):

kraken2 --db PATH/kraken/plant_original --threads 8 --confidence 0.1 --report PATH/SB0new_report.txt PATH/SB0.fastq.gz --report-zero-counts --output PATH/SB0new_taxa.txt

However, the new kraken2 report is exactly the same as the old one and it did not find any hit to these new added taxa.

Previously, I did a blastn alignment of these sample’s reads with these exact same taxa (I created a small db with alnus_glutinosa_GCA_003254965.1.fna, carpinus_fangiana_GCA_006937295.1.fna etc..) and blastn could find some hits. So I was expecting that Kraken2 would have found these matches too.

Are these steps wrong or am I making some conceptual mistake?

I appreciate your help.

Thanks

[davidkim8]

I think I'm running into similar problem. Have you found solution yet? I'm thinking I have to delete all the .k2d files and remake everything with new database (which will take a long time I think).

Answer 1

[Mark]

You can not add genomes to an existing, precomputed kraken db. You have to download the plant genomes, add the new genomes then recreate the db.

[valentinavan]

Hi Mark,

thanks for replying. Is it not what I have done? I have downloaded the genomes and added them to the db. Can you please be more specific so that I can understand. Thanks

[pbioinf]

Edit: Mark is right. You added new files to an existing library. Do you have the fasta files of the initial plant database?

[valentinavan]

Hi,

Thanks for your help. I made custom database (from scratches), I tested Kraken2 with a sample and I got matches, however, when I created the db it also produced an unmapped.txt file with a long list of accession numbers. I am not sure how to treat this...

Please see this other post: Custom Kraken2 db