Macrophage responses across infection with two L. panamensis strains and drug treatment

This repository contains the recipe, data, etc. used to perform the various analyses when examining human macrophages infected (or not), drug treated (or not) with two zymodemes of Leishmania panamensis.

Caveats

In two potentially important ways this repository does not completely meet my definition of a reproducible image:

1. The recipe explicitly uses the already-built image of hpgltools in the parent recipe: https://github.com/elsayed-lab/hpgltools_singularity This image was digitally signed (by me) and uploaded to the singularity hub: https://cloud.sylabs.io/library/abelew/hpgltools/hpgltools.sif Thus, I am not sure if another person can prove that I didn't do some shenanigans inside that parent image. However, I do clone the repository into /hpgltools within the resulting container, so one may trivially go into it and look around and see the state of the codebase.
2. In earlier revisions of this and its sibling images, I use an explicit git commit of hpgltools to make absolutely certain that the resulting image used a known state of that repository. Given that I am using the prebuilt image in #1, that is no longer true.
3. I didn't add the actual data yet to the repository. Once the manuscript is (close to?) finished, I will perform the git add of the 'data/' tarball.
4. You must (for the moment) assume that I treated the raw data in a sane fashion. Once we have SRA IDs, I will create a recipe/image which may be used to explicitly create all the data used here, though that image will possibly not include the full installation of fastp,fastqc,trimomatic,samtools,freebayes,salmon,hisat,etc... I am not sure how I want to handle that yet.

If one is willing to overlook these two caveats, then I think this recipe nicely meets the definition of a reproducible codebase. It is also quite trivial to fix the above at the cost of a (much) longer build time:

1. Go into the recipe.yml and change the second line to 'From: debian:stable' instead of the reference to library:// (oh, this image is not using that sylabs image yet, so I guess nevermind?)
2. Change the local/bin/setup_hpgltools.sh so that the variable 'commit' points to the git commit ID of interest.

Installation/Usage

1. Clone the repository.
2. run 'make tmrc2_macr.sif' or 'sudo singularity build tmrc2_macr.sif tmrc2_macr.yml'
3. wait a while.
4. The resulting .sif file may be run as a standalone executable. It currently assumes the user set the SINGULARITY_BIND environment variable to something like 'SINGULARITY_BIND=.:/output' so that it may write its set of rendered html reports to the current working directory. Assuming that is true, it will create a new tree with a prefix of '$(date +%Y%M)' and put them there. It also copies the input data, sample sheets, etc there for one to play with.

Failures observed when moving to bioc 3.18:

A change in devtools caused me to change how I install the various dependencies/prerequisites. This in theory should be more robust, but in practice is a big mess. It turns out that there is a significant population of packages which worked in 3.17 but not in 3.18. As a result I am now handling this in two passes: one where I allow devtools to install whatever version it chooses for each package -- these tend to be bioc 3.17. I then take a second pass via BiocManager::install() to update some/many to the 3.18 version.