docs: Fix units configuration docs for bam files

[fxwiegand]

Just a quick PR to update the docs for the recent changes from #94.

[dlaehnemann]

LGTM. One thing we might think about, is making the units.tsv also work with just one of those options' columns present. That way, we wouldn't have to always drag around a bunch of empty columns. But this might entail changing quite a bit of code whenever these columns are called. What do you think?

[fxwiegand]

LGTM. One thing we might think about, is making the units.tsv also work with just one of those options' columns present. That way, we wouldn't have to always drag around a bunch of empty columns. But this might entail changing quite a bit of code whenever these columns are called. What do you think?

I absolutely see the point you're making. We would need to make the pandas calls on these lines a more robust so they can handle when the bam columns are not present:

rna-seq-kallisto-sleuth/workflow/rules/common.smk

Lines 93 to 112 in 8d9773e

	return fq2_not_present and bam_paired_not_present
	def get_fastqs(wildcards):
	"""Get raw FASTQ files from unit sheet."""
	if not pd.isnull(units.loc[(wildcards.sample, wildcards.unit), "bam_single"]):
	return f"results/fastq/{wildcards.sample}-{wildcards.unit}.fq.gz"
	elif not pd.isnull(units.loc[(wildcards.sample, wildcards.unit), "bam_paired"]):
	fqfrombam1 = f"results/fastq/{wildcards.sample}-{wildcards.unit}.1.fq.gz"
	fqfrombam2 = f"results/fastq/{wildcards.sample}-{wildcards.unit}.2.fq.gz"
	return [fqfrombam1, fqfrombam2]
	elif is_single_end(wildcards.sample, wildcards.unit):
	return units.loc[(wildcards.sample, wildcards.unit), "fq1"]
	else:
	u = units.loc[(wildcards.sample, wildcards.unit), ["fq1", "fq2"]].dropna()
	return [f"{u.fq1}", f"{u.fq2}"]
	def get_all_fastqs(wildcards):
	for item in units[["sample", "unit"]].itertuples():

Besides that these lookup calls from snakemake would cause trouble. I am not sure how they handle missing columns.

rna-seq-kallisto-sleuth/workflow/rules/bam.smk

Line 6 in 8d9773e

cols="bam_paired",

rna-seq-kallisto-sleuth/workflow/rules/bam.smk

Line 25 in 8d9773e

cols="bam_single",

[dlaehnemann]

You can nowadays offer a default="" argument in the lookup() function, to capture those: https://snakemake.readthedocs.io/en/stable/snakefiles/rules.html#the-lookup-function. We just have to enforce the latest Snakemake version, then. Would be willing to create a new PR referencing this discussion? Then we can continue discussing there, as well.

[fxwiegand]

You can nowadays offer a default="" argument in the lookup() function, to capture those: https://snakemake.readthedocs.io/en/stable/snakefiles/rules.html#the-lookup-function. We just have to enforce the latest Snakemake version, then. Would be willing to create a new PR referencing this discussion? Then we can continue discussing there, as well.

Yeah i will give it shot and we will discuss further in the new PR! This would also be quite nice for saving the versioning because now i think about it #94 was indeed a breaking change since the config needs the new bam columns for the workflow to run.

[dlaehnemann]

Yeah, I think you are right. But I have honestly not payed enough attention to what's a breaking change in these workflows in the past. But we should. This regularly trips up users (us included). 😅

[fxwiegand]

Yeah, I think you are right. But I have honestly not payed enough attention to what's a breaking change in these workflows in the past. But we should. This regularly trips up users (us included). 😅

Yep indeed - @Addimator #105 will let you remove the bam columns from your config again.