https://nf-co.re logo
Join Slack
Powered by
Patch release 1.4.1 fixing the problem with paths ...
# metatdenovo
d
Patch release 1.4.1 fixing the problem with paths not mounted by Singularity: github.com/nf-core/metatdenovo/…/1.4.1. To run this release:
Copy code
nextflow pull nf-core/metatdenovo -r 1.4.1
nextflow run nf-core/metatdenovo -r 1.4.1 -profile <...> --outdir  <...> <other params
b
Thanks a lot, just submitted 👍
d
Keep me posted!
👍 1
b
Hi Daniel, short update from my side. Both bbnorm and the megahit co-assembly finished without problems using v 1.4.1. Using now the assembly as input and try to get the rest of the pipeline (currently prokka) to run through.
🥳 2
Quick question: how much memory and CPUs did you allocate to BBNORM_INDEX? My job fails using 192.GB because of OOM. Tried to allocate 256.GB and cpus = 8, but apparently I can’t specify these node configurations (perhaps due to maintenance?).
d
I had to go to 900 GiB and 96 cores! Here's the full config block:
Copy code
withName: BBMAP_BBNORM {
        ext.args = { [
            "target=${params.bbnorm_target}",
            "min=${params.bbnorm_min}",
            "prefilter=t"
        ].join(' ').trim() }
        cpus   = { 96 }
        memory = { 900.GB }
        time   = { 30.d }
    }
The 
prefilter=t
is meant to make it more accurate when you have too little memory.
b
Thanks a lot. It seems like this step also finished. I am currently trying to run PROKKA without success. It seem to be connected to the blast step: .command.err
Copy code
[13:33:21] Found 9682 CDS
[13:33:21] Connecting features back to sequences
[13:33:21] Not using genus-specific database. Try --usegenus to enable it.
[13:33:21] Annotating CDS, please be patient.
[13:33:21] Will use 20 CPUs for similarity searching.
[13:33:22] There are still 9682 unannotated CDS left (started with 9682)
[13:33:22] Will use blast to search against /opt/conda/db/kingdom/Bacteria/IS with 20 CPUs
[13:33:22] Running: cat megahit_assembly\.contigs\.22\/megahit_assembly\.contigs\.22\.IS\.tmp\.42\.faa | parallel --gnu --plain -j 20 --block 50254 --recstart '>' --pipe blastp -query - -db /opt/conda/db/kingdom/Bacteria/IS -evalue 1e-30 -qcov_hsp_perc 90 -num_threads 1 -num_descriptions 1 -num_alignments 1 -seg no > megahit_assembly\.contigs\.22\/megahit_assembly\.contigs\.22\.IS\.tmp\.42\.blast 2> /dev/null
[13:33:22] Could not run command: cat megahit_assembly\.contigs\.22\/megahit_assembly\.contigs\.22\.IS\.tmp\.42\.faa | parallel --gnu --plain -j 20 --block 50254 --recstart '>' --pipe blastp -query - -db /opt/conda/db/kingdom/Bacteria/IS -evalue 1e-30 -qcov_hsp_perc 90 -num_threads 1 -num_descriptions 1 -num_alignments 1 -seg no > megahit_assembly\.contigs\.22\/megahit_assembly\.contigs\.22\.IS\.tmp\.42\.blast 2> /dev/null
d
Can you upload the 
.command.*
files from that? The problem with the above log -- which will remain when you send the full files -- is that standard error is sent to 
/dev/null
, the bit bucket. But, we'll see what Claude can make out of the files first. It might have follow-up questions
b
.command.err,.command.log,.exitcode,.command.run,.command.sh
.command.err.command.sh.command.run.command.log
👍 1
d
Same problem as before but with a different module. Add this to the top of your config: 
singularity.runOptions = '-B "$TMPDIR"'
and restart
👍 1
b
It seems to complete each job now without an exit status 🤞
👍 1
hs_err_pid68.log
hs_err_pid68.log
That would be the memory for BBMAP_INDEX, whats your setting there?
d
Copy code
withName: BBMAP_INDEX {
        cpus   = { 48 }
        memory = { 400.GB }
        time   = { 3.d }
    }
    withName: BBMAP_ALIGN {
        cpus   = { 48 }
        memory = { 400.GB }
        time   = { 3.d }
    }
b
Thanks a lot, will try it with 400.GB
2 Views
Open in Slack
PreviousNext
Powered by