Howdy! I'm running the pipeline with a non-standar...
# taxprofiler
m
Howdy! I'm running the pipeline with a non-standard but previously published database (https://ezmeta.unige.ch/CMMG/) and trying to do bracken as well. I have run_bracken in my command and it is registering but the actual bracken command isn't running. I downloaded all the files in Krakendb/ and taxonomy/. What else do I need to do to get Bracken to actually run?
j
What does your database.csv look like?
m
tool,db_name,db_params,db_path kraken2,db1,--threads 6 --classified-out classified#.fastq --unclassified-out unclassified#.fastq --paired --gzip-compressed ,/scratch/mjustice/k2_standard_20250714.tar.gz bracken,db1,;--threads 6,/scratch/mjustice/k2_standard_20250714.tar.gz
Those are the full contents. It actually kicked off on my last try but exceeded memory limits so I'm going to attempt to increase the memory and try again
Hey, James! I ran the pipeline with bracken and I can see some related files in the kraken2 folder (sample.bracken.kraken2.report.txt). However, the pipeline skips the other bracken steps saying it can't find the .tsv file needed. No bracken folder is made in the results section. Have you ever seen this before?
Update - I had just left out the -r parameter. Added it and got the pipeline to finish with the Standard kraken2 db. Now trying to get the cmmg database run to work with bracken. It was having issues finding the files so I've kicked off a whole new instance of the pipeline to see if that fixes it.
j
I would need to see the full error messages to see
Does your cmmg database have the addition
kmer
files inside it?
m
Ahh. Ok. It does have several, but my read length is 300 and there is no 300mers file.
I'll do some digging to find the right bracken-build command to put that together, but if you have it off hand that'd be helpful!
I ended up just running it with the 250mer file!
j
Too late, but I have a whole pipeline for building databases 😬 #C062H82UCRG (although there is a sticky bug atm with it :/)
And 250 seems about right, you want your kmers to be shorter than your reads as a rough rule anyway