HiSAT2 with Python 3k

[ewels]

Hi all,

https://bioconda.github.io/contributor/guidelines.html states that "packaging is hard" - I agree! 😆 But one of the examples is:

hisat2 runs 2to3 to make it Python 3 compatible, and copies over individual scripts to the bin dir

Sure enough, the build recipe does run 2to3:

bioconda-recipes/recipes/hisat2/build.sh

Lines 5 to 20 in 9062056

	if [ $PY3K -eq 1 ]
	then
	2to3 --write \
	hisat2_extract_exons.py \
	hisat2_extract_snps_haplotypes_UCSC.py \
	hisat2_extract_splice_sites.py \
	hisat2-build hisat2-inspect \
	hisat2_simulate_reads.py \
	hisat2_extract_snps_haplotypes_VCF.py \
	hisatgenotype_build_genome.py \
	hisatgenotype_extract_reads.py \
	hisatgenotype_extract_vars.py \
	hisatgenotype_hla_cyp.py \
	hisatgenotype_locus.py \
	hisatgenotype.py
	fi

However, the meta.yml file then skips python 3k builds:

bioconda-recipes/recipes/hisat2/meta.yaml

Line 13 in 9062056

skip: True # [not py27]

So despite making the code python 3 compatible, HiSAT2 ends up being available for Python 2 only.

This skipped build was added by @pvanheus and @bgruening last year in #9973 - can you guys remember why? Do you think it's possible to add python 3 support back again?

Phil

[bgruening]

@pvanheus can you please comment here?

[ewels]

What’s the best course of action here? Shall I prep a PR enabling Py3k builds again?

[ewels]

I updated the build.sh file to run 2to3 conversion on all HiSAT2 python scripts (several were missing) and allowed it to build for all python versions.

The CI tests seemed to pass. Testing functionality now, but I think that this may be fixed now..? 🤞

[ewels]

It's a slightly obscure test, but the nf-core/methylseq pipeline now runs with bismark using HiSAT2 on a python 3k environment inside a docker container! 😆

https://travis-ci.com/nf-core/methylseq/jobs/263630208

I'm happy with this, so will close this issue now.