get ready for submission to Bioconductor

These are the tasks that we should finish before submission (feel free to edit and add or tick off):

[x] check if package name is available (available::available("monaLisa")), rename package if needed
[x] Michael: switch from parallel::mclapply to BiocParallel::bplapply in order to also support Windows
[x] Michael: top-level wrapper function: calcBinnedMotifEnr(, method = c("R", "Homer")) -> calls runHomer() or get_binned_motif_enrichment()
[x] rename runHomer() and get_binned_motif_enrichment() to calcBinnedMotifEnrHomer() and calcBinnedMotifEnrR()?
[x] for calcBinnedMotifEnr(), make both method = c("R", "Homer") consistent (e.g. naming of rows in the returned SE, names used in the plots)
[x] for runHomer(): also return the count matrices sumForegroundWgtWithHits and sumBackgroundWgtWithHits
[x] use meaningful variable names (e.g. motifCounts instead of D); use them consistently in similar functions (e.g. runHomer() and get_binned_motif_enrichment())
[x] Problem for log2enr assay in runHomer(): Homer sometimes incorrectly reports the total numbers of sequences in its text output - for example it reported a higher number of total foreground sequences (3006) than had been given as input (3000) in one case. This usually does not result in large differences and only affects the log2enr assay (colnames(D)); p-values are extracted directly and Homer internally uses the correct totals to calculate these. Getting the total from the 100% calculated from number of sequences with hits and percentage of sequences with hits is even more noisy and not an option. Should we document this? Is it reproducible?
[x] for calcBinnedMotifEnr(), only accept pwmL for method = "R" and only accept motif file for method = "Homer" (add examples to vignette of how to transform one to the other)
[x] Michael: for calcBinnedMotifEnr(), naming of assays (pearsonResid, negLog10P, negLog10Padj instead of enr, p, FDR?)
[x] Michael: testing of method = "Homer": switch to synthetic data so that tests also run on GHA (currently: missing genomedir)
[x] Michael: use R-based instead of Homer-based binned motif enrichment analysis in vignette
[x] Michael: replace use of hg19 TxDb and BSgenome with mm10 (reduce dependencies)
[x] Michael: use graphics::strwidth to define ylim in plotSelectionProb()
[x] Michael: change color of FDR (gray?)
[x] add ... to plotMotifHeatmaps() (passed to ComplexHeatmaps::Heatmap)
[x] Michael: implement issue #35 (background definitions; 4 modes)
[x] Lukas: implement #29 (no enrichment in randomly selected/binned regions)
[x] add line to vignette how to remove central bin (use: seplot <- se[, -attr(b, "bin0")]) in plotMotifHeatmaps. (use getZeroBin), select motifs (excluding zero Bin) and plot all bins.
[x] Michael: for binned motif enrichment: parameterize pseudocount in motif enrichment calculations?
[x] Michael: for binned motif enrichment: parameterize the multiple testing correction method?
[x] Michael: for binned motif enrichment: parameterize GC bin breaks (default to HOMER)
[x] Dania: for binned motif enrichment: make test="fisher" the default in .calcMotifEnrichment? yes. We need to compare more to the Homer run with the Hypergeometric test, if we get matching results.
[x] Dania: for binned motif enrichment: add motifs without hits (in both fore- and background) to assays as NA values? (currently they are reported with a p-value of 1 / log2enr of 0)
[x] Michael: for stability selection SE: add colData column with "area under the selection curve"
[x] Dania: For stability selection plotSelectionProb() function: make barplots horizontal and add motif seqlogos on the side? (also add AUC as row annotation?)
[x] Dania: revisit the example in the vignette for stabsel
[x] Dania: stabsel in vignette: mention case that 2 highly correlation predictors (example same one twice), may not be selected at all
[x] Dania: revisit the example dataset for stabsel and randomly subsample a number of peaks rather than select with logFC cutoffs (user may think this is what they should do otherwise)
[x] add expTF assay for all binned enrichment functions
[x] consistent way of calculating pearson resid (after Lukas has had a look)
[ ] vignette: add section about issues with similar motifs (cluster motifs by motifSImilarity) or more generally colinear predictors in stability selection. Discuss how to detect such cases, and the issues arising from it, and maybe possible work-arounds (selecting one representative per correlation group?)
[x] Dania: vignette: add illustration of creating PWM from txt file and how to convert between the 2?
[x] ? adjust dumpJaspar help page, to indicate it's using PFMs to get PPMs and not using PWMs as stated?
[x] vignette: clean up use of references (literature citations, section links, Bioc/CRAN packages)
[x] Lukas: vignette: add example (data/analysis) for DNAseI TKO-wt
[x] move experimental stuff into seperate branch (e.g. some k-mer based analyses? sample_random_regions()?)
[x] vignette / tests: make sure they also run if Homer is not available
[x] merge Homers_better_looking_twin into master
[x] change version number to 0.99.0
[x] add maintainer field to DESCRIPTION
[x] check runtime for 'R CMD check --no-build-vignettes' is less than 10 minutes (6-7min on xenon6)
[x] check individual files <= 5MB (checked on Sept 10; installed size is ~7MB)
[x] in binned motif enrichment: add warning when seqlengths are different, and warn the user that there is a length adjustment being implemented, and that the user should consider maybe changing the input seqs to have the same lengths. #14 f27b0828
[x] Dania: make separate vignette for stability selection.
[x] from homer output, create the contingency table to calculate the pearson residual the same way as in the rest of monaLisa
[x] Dania: in stability selection: remove supressWarnings() in stabSel function, why is it there.
[x] Update inst/extdata/se.rds so that it represents the most up-to-date return value of calcBinnedMotifEnr() (with expexted count assay etc). Rename in a more descriptive way.
[x] Add documentation of data generation in inst/scripts/
[x] Don't use wrapper motif enrichment function in vignette/examples.
[x] Dania: remove length normalization

Info for submission process is available here: http://www.bioconductor.org/developers/package-submission/ Package guidelines are outlined here: http://www.bioconductor.org/developers/package-guidelines/

fmicompbio / monaLisa

get ready for submission to Bioconductor #36