Pipeline unable to recognise samples processed across multiple lanes

[jma1991]

Description of the bug

I've identified a potential issue in the recent pipeline release (v2.5.0). It seems the groupTuple command is executed twice during the input channel creation and branching of FASTQ files. As a result, the pipeline is unable to recognise samples processed across multiple lanes, due to an additional layer of file nesting. See here:

methylseq/workflows/methylseq.nf

Lines 98 to 105 in 66c6138

	.groupTuple()
	.map {
	meta, fastq ->
	def meta_clone = meta.clone()
	meta_clone.id = meta_clone.id.split('_')[0..-2].join('_')
	[ meta_clone, fastq ]
	}
	.groupTuple(by: [0])

Command used and terminal output

No response

Relevant files

No response

System information

No response

[mz448]

This believe this is an issue with the samplesheet.csv info

Discussion reference

Follow @FelixKrueger and @bioinfoMMS discussion in the slack channel -> conversation

How to "solve" it:

I was able to run the pipeline using bismark by adding an underscore "_" inside the name of the sample (in column 1) in the samplesheet.csv
e.g. ( use sample1_rep1 instead of sample1)
Make sure you use 4 header columns instead of 3 being the last genome. (this isn't very clear because the current documentation at https://nf-co.re/methylseq does not mention it! But Felix says it in the conversation

e.g.:

# use this:
sample, fastq_1, fastq_2, genome
sample1_rep1, bla/sample1_R1.fastq.gz, bla/sample1_R2.fastq.gz,
sample2_rep1, bla/sample2_R1.fastq.gz, bla/sample2_R2.fastq.gz,

# instead of:
sample,fastq_1,fastq_2
sample1, bla/sample1_R1.fastq.gz, bla/sample1_R2.fastq.gz
sample2, bla/sample2_R1.fastq.gz, bla/sample2_R2.fastq.gz

To add multiple lanes of the same sample, repeat the name of the sample, and they will merge during the processing.

[wkang0]

This is bug in the instruction instead of in the code. To make one sample in different lanes, the sample sheet should look like this:

sample1_REP1,fq1.gz,fq2.gz
sample1_REP2,fq11,ga,fq12.gz