Add other packages for xqtl-protocol

[danielnachun]

We are now at the point where we can add the other packages used in various parts of the xQTL protocol to our setup. Some of the packages we originally used in our containers may no longer be relevant to the current version of the protocol, so we can omit those from our pixi setup.

CLI tools

• apex - QTL mapping tool - maybe not relevant given our focus on individual-level SuSiE?
• bamtools - unclear if this is needed if we also have samtools
• dap-g - unclear if this is needed given that we are using SuSiE for fine mapping
• deeptools - tool for working with multiple types of NGS files
• ensembl-vep - will be needed for annotation pipeline for rare variants
• fastenloc - unclear if this is needed now that we have pecotmr
• fastp - tool to work with FASTQs
• finemap - unclear if this is needed given that we are using SuSiE for fine mapping (proprietary, Intel only)
• gcta - toolkit for statistical genetics - I haven't fully gotten this to build on Apple Silicon yet
• gffread - tool to work with GTF/GFF files
• gtex-pipeline-gene_model
• gtex-pipeline-qtl
• gtex-pipeline-rnaseq
• kallisto - unclear if this is needed unless we using its quantifications
• king - kinship estimation tool
• ldstore - tool for storing LD (proprietary, Intel only)
• metal - meta-analysis tool
• multiqc - QC reporting tool
• polyfun - tool for functionally informed fine mapping - has a huge dependency tree
• quantas - tool for working with splicing - Ru is using a fork for this as well
• regtools - tool for quantifying splicing - needed as input for LeafCutter
• rsem - isoform quantification tool - not sure if we actually using RSEM quantifications for anything
• samtools - tool for working SAM/BAM files
• star - aligner for RNA-seq
• suppa - splicing quantification tool
• torus - heritability enrichment tool

R packages

• r-bedtoolsr

Python packages

• mofapy2
• multipy
• pybigwig
• python-ldstore
• qvalue

[gaow]

Thanks @danielnachun I think there are two situations here, 1) those tools that we don't use for sure and 2) those that we still include as part of our protocol but never really interpreted them (bandwidth mainly)

For those we don't use:

• apex - QTL mapping tool - maybe not relevant given our focus on individual-level SuSiE?
• dap-g - unclear if this is needed given that we are using SuSiE for fine mapping
• fastenloc - unclear if this is needed now that we have pecotmr
• finemap - unclear if this is needed given that we are using SuSiE for fine mapping (proprietary, Intel only)
• metal - meta-analysis tool
• suppa - splicing quantification tool
• torus - heritability enrichment tool

For those still nice to keep but not used right now:

• ldstore - tool for storing LD (proprietary, Intel only)
• rsem - isoform quantification tool - not sure if we actually using RSEM quantifications for anything
• python-ldstore

Those that I'm not sure of:

• kallisto - unclear if this is needed unless we using its quantifications
  • I also don't know where this is used. Need to clarify with Frank
• polyfun - tool for functionally informed fine mapping - has a huge dependency tree
  **For this one we never really run it. We have all outputs from S-LDSC though (and we need S-LDSC). I wonder after installing S-LDSC and SuSiE we should automatically have all polyfun dependencies? What extra was it doing? **

Everything else is what we actively use.

[danielnachun]

All of the missing packages except kallisto and polyfun have now been added to the setup script, and should have a negligible impact on the size of the installation. The only change I'm hoping to make in the future is for gcta to support Apple Silicon.

kallisto is very small so it's just a question of whether we actually need it or not - hopefully Frank can figure that out soon.

For polyfun we'll have to understand what we'd use it for and we may want it to be optional because it has so many dependencies.