Replace featureCounts with Salmon quant

[drpatelh]

• Replace --aligner star with --aligner star_salmon

• Remove HISAT2 quantification

• Add Groovy functions to check for --genome existence and to get reference files

• Moved all genome related file processing and indices creation to prepare_genome sub-workflow

• Fixed Add decoy-aware indexing for salmon #515 - Add decoy-aware indexing for salmon

• Fixed Significantly different versions of STAR in star_rsem (2.7.6a) and star (2.6.1d) #498 - Significantly different versions of STAR in star_rsem (2.7.6a) and star (2.6.1d)

• Partially fixed Use of salmon counts for DESeq2 #499 - Use of salmon counts for DESeq2

• Parameter changes:

Old parameter	New parameter
--fc_extra_attributes	--gtf_extra_attributes
--fc_group_features	--gtf_group_features
--fc_count_type	--gtf_count_type
--fc_group_features_type	--gtf_group_features_type
--skip_featurecounts	-

[apeltzer]

LGTM, maybe the checking that NXF 20.10 is used but otherwise good

[drpatelh]

Thanks @apeltzer @grst ! Will bump Nextflow version in another PR because these changes don't require the latest version.

[harish0201]

Not an issue, just a doubt

Hi, is there any reason for removing quantification from HiSAT2 alignments?

[drpatelh]

Hi @harish0201 . I wrote a brief description for this in the CHANGELOG (see links to papers there too) and a footnote in the main README at the end of this section.

But to summarise, HISAT2 only generates alignments relative to the genome and not the transcriptome so we can't use tools such as salmon quant (yet) for quantification (see Note). The quantification methods used by featureCounts aren't suited for expression level quantification especially when trying to factor in multiple isoforms / pseudogenes etc which is why it was removed as a downstream quantification option.