hostremoval_input_fastq has insufficient memory

[ivelsko]

Check Documentation

I have checked the following places for your error:

• [x] nf-core website: troubleshooting
• [x] nf-core/eager pipeline documentation

Description of the bug

hostremoval_input_fastq fails on larger samples b/c it runs out of memory

Steps to reproduce

Steps to reproduce the behaviour:

1. Command line:

   nextflow run nf-core/eager \
   -r 2.3.5 \
   -profile eva,archgen,big_data \
   --outdir /mnt/archgen/microbiome_calculus/abpCapture/03-preprocessing/set1_set3/eager2 \
   -work-dir /mnt/archgen/microbiome_calculus/abpCapture/03-preprocessing/set1_set3/work \
   --input /mnt/archgen/microbiome_calculus/abpCapture/03-preprocessing/set1_set3/abpCap_set1_set3_eager_input.tsv \
   --complexity_filter_poly_g \
   --fasta /mnt/archgen/Reference_Genomes/Human/HG19/hg19_complete.fasta \
   --seq_dict /mnt/projects1/Reference_Genomes/Human/HG19/hg19_complete.dict \
   --bwa_index /mnt/archgen/Reference_Genomes/Human/HG19/ \
   --bwaalnn 0.02 \
   --bwaalnl 1024 \
   --hostremoval_input_fastq \
   --hostremoval_mode remove \
   --run_bam_filtering \
   --bam_unmapped_type fastq \
   --skip_damage_calculation \
   --skip_qualimap \
   --email irina_marie_velsko@eva.mpg.de \
   -name abpCap_set13 \
   -with-tower

2. See error:

   
   Error executing process > 'hostremoval_input_fastq (BSH001.A0101.SG1)'

Caused by: Process hostremoval_input_fastq (BSH001.A0101.SG1) terminated with an error exit status (1)

Command executed:

samtools index BSH001.A0101.SG1_PE.mapped.bam extract_map_reads.py BSH001.A0101.SG1_PE.mapped.bam BSH001.A0101.SG1_R1_lanemerged.fq.gz -rev BSH001.A0101.SG1_R2_lanemerged.fq.gz -m remove -of BSH001.A0101.SG1_PE.mapped.hostremoved.fwd.fq.gz -or BSH001.A0101.SG1_PE.mapped.hostremove d.rev.fq.gz -p 1

Command exit status: 1

Command output:

• Extracting mapped reads from BSH001.A0101.SG1_PE.mapped.bam
• Parsing forward fq file BSH001.A0101.SG1_R1_lanemerged.fq.gz

Command error: Traceback (most recent call last): File "/home/irina_marie_velsko/.nextflow/assets/nf-core/eager/bin/extract_map_reads.py", line 270, in File "/home/irina_marie_velsko/.nextflow/assets/nf-core/eager/bin/extract_map_reads.py", line 147, in parse_fq File "/home/irina_marie_velsko/.nextflow/assets/nf-core/eager/bin/extract_map_reads.py", line 120, in get_fq_reads File "/opt/conda/envs/nf-core-eager-2.3.5/lib/python3.7/site-packages/Bio/SeqIO/QualityIO.py", line 933, in FastqGeneralIterator seq_string = handle_readline().rstrip() File "/opt/conda/envs/nf-core-eager-2.3.5/lib/python3.7/site-packages/xopen/init.py", line 268, in readline return self._file.readline(*args) File "/opt/conda/envs/nf-core-eager-2.3.5/lib/python3.7/codecs.py", line 322, in decode (result, consumed) = self._buffer_decode(data, self.errors, final) MemoryError

Work dir: /mnt/archgen/microbiome_calculus/abpCapture/03-preprocessing/set1_set3/work/e4/663badafbd377d9291bdb211a98525

Tip: when you have fixed the problem you can continue the execution adding the option -resume to the run command line -[nf-core/eager] Pipeline completed with errors-

## Expected behaviour

There should be enough memory for the host-mapped reads to be removed from the input files and the resulting host_removed fastq files to be written for both forward and reverse reads. 

## Log files

Have you provided the following extra information/files:

- [x] The command used to run the pipeline
- [ ] The `.nextflow.log` file <!-- this is a hidden file in the directory where you launched the pipeline -->
- [x] The exact error: see above

## System

- Hardware: HPC
- Executor: sge
- OS: Linux
- Version: Ubuntu 20.04.3 LTS

## Nextflow Installation

- Version: 20.10.0 build 5430

## Container engine

- Engine: Singularity
- version: <!-- [e.g. 1.0.0] -->
- Image tag:  nfcore/eager:2.3.5 

## Additional context

I tried to increase the memory from 32GB by adjusting the lines 

$ -l h_rss=184320M,mem_free=184320M

$ -S /bin/bash -j y -o output.log -l h_vmem=180G,virtual_free=180G

in the `.command.run` file. It ran with 180GB as written above, but the `qacct` record says `maxvmem      120.745GB`.

[ivelsko]

@maxibor for you, but I can't seem to assign you

[jfy133]

@maxibor more specifically it seems this ends up using a ridiculous amount of memory, so I think it would require optimisation on part of the script.

[jfy133]

@maxibor also posted the following on slack:

https://github.com/TOAST-sandbox/podPeople https://github.com/sandberg-lab/dataprivacy (a.k.a. BAMboozle)

Note: these tools are probably indeed more robust, but we need to consider that it might be unwanted to just 'replace' reads or variants with ref. genome ones. If someone wanted to reanalyse e.g. calculus for human DNA, they may not realise they are lookin at 'fake' sequence.

This is sort of 'tampering' with the FASTQ file in a misleading sense. Therefore I would still rather have a NNN replacement or entire removal (will need to check if the tools support this).

[jfy133]

Done in 2.4.5!