nf-core / mag

Assembly and binning of metagenomes
https://nf-co.re/mag
MIT License
209 stars 107 forks source link

Skipping NFCORE_MAG:MAG:BOWTIE2_PHIX_REMOVAL_ALIGN #619

Closed aboddicker closed 4 months ago

aboddicker commented 4 months ago

Description of feature

I'd like to skip the "NFCORE_MAG:MAG:BOWTIE2_PHIX_REMOVAL_ALIGN" process. My reads are already removed of PhiX and I find this step to be very time consuming, even for lower numbers of reads (<100 million). Some jobs timeout after 6 hours (>100M reads). I've attempted to changer parameters in the config to "--very-fast" or increasing threads "-p" but the arguments are not recognized. I have not found a way to skip this process altogether. I am using Seqera Cloud to launch jobs. Thanks!

jfy133 commented 4 months ago

Related: https://github.com/nf-core/mag/issues/455

d4straub commented 4 months ago

Why not use --keep_phix ? This should not run the process you are disliking, see https://github.com/nf-core/mag/blob/a50117e4ddc681b8ec5535fc265c66df45e38d30/workflows/mag.nf#L275-L287

jfy133 commented 4 months ago

Yeah I forgot about that (I guess possibly the reason why @aboddicker also didn't recognise it, I find it rather unintuitive :sweat_smile: ) - please let us know if that works @aboddicker ! Then we close that and refer to the other issue for making the method of opting in/out more consistent

aboddicker commented 4 months ago

Thanks so much! Yes, this is a great solution. Selecting --keep_phix bypasses the alignment and proceeds simultaneously with FastQC and into assembly upon launch.

aboddicker commented 4 months ago

Closing ticket. --keep_phix skips initial Bowtie2 alignment of reads to PhiX reference.