drop go dependency ?

[cmdoret]

Hi @ABignaud, I noticed you are using dnaglider in your smk workflow 😁
Unfortunately I did not make it into a conda package, which is a shame because Go is only required to compile it (and not run it).

As I don't have time to make a go pkg now, maybe a more convenient alternative for you would be seqkit, another go package which is on bioconda.

Here:

transcription_T7_analysis/rules/04_ecoli_matrix_analysis.smk

Lines 15 to 22 in 8de68b1

	go get -u github.com/cmdoret/dnaglider/dnaglider
	dnaglider \
	-fasta {input.fasta} \
	-fields GC \
	-out {output.gc} \
	-stride 100 \
	-threads {threads} \
	-window 500

You could have something like this instead:

seqkit sliding --step 100 --window 500 W303_SGD_2015_JRIU00000000.fa.gz | seqkit fx2tab --gc | cut -f1,4

gi|696447993|gb|JRIU01000414.1|_sliding:901-1400	38.80
gi|696447993|gb|JRIU01000414.1|_sliding:1001-1500	38.80
gi|696447993|gb|JRIU01000414.1|_sliding:1101-1600	38.60
gi|696447993|gb|JRIU01000414.1|_sliding:1201-1700	38.60
gi|696447993|gb|JRIU01000414.1|_sliding:1301-1800	38.80
gi|696447993|gb|JRIU01000414.1|_sliding:1401-1900	39.80
gi|696447993|gb|JRIU01000414.1|_sliding:1501-2000	40.00
gi|696447993|gb|JRIU01000414.1|_sliding:1601-2100	39.80

This would allow you to just add seqkit to the conda env and drop go from dependencies

[ABignaud]

Yes I agree it would have been better to avoid the dependency issues. I just start to do the workflow to compile the analysis pipelines I used to generates the figures for the paper. I used the output of dnaglider to generate the GC file (outside the workflow at the beginning) used in the analysis so that's why I used it in the rule even if the dependency is painful.

[cmdoret]

Makes sense, I added an issue on cmdoret/dnaglider#4 (comment) I am not sure when I will have time to make a conda package, but providing a dockerfile should be fairly simple (and better than nothing).