todo.org

bcbioSingleCell

Development

Need to address this note now popping up in build checks.

Note: found 88 marked UTF-8 strings

Need to ensure R Markdown renders correctly without goalie issues.

Require valid names in all slots.

Add monocle celldataset coercion method.

Need to run BFG on the repo and remove old example R data.

`bcbioSingleCell()`: Explain genome annotation priority in better detail, following the logic defined in `bcbioRNASeq()`.

Improve support for transcript-level counts?

Should we even allow this at this point?

Vignette using example bcbio data and 10X pbmc data.

Look into BarcodeInflectionsPlot, now in Seurat 3.

pointillism

Move markers out of Google Sheets and into the package, as simple CSV files.

Use pseudobulk approach for DE with sample replicates

`plotFeature()`: Add `pointsAsNumbers` argument support.

`diffExp()`: Add internal support for accessing design with `design()` generic.

`KnownMarkers` S4: switch from DataFrame to SplitDataFrameList inheritance?

Improve SeuratMarkers class.

This needs to error if the input data.frame contains `cluster` column. Consider only using `SeuratMarkers` as a single generator but returning `SeuratMarkers` or `SeuratMarkersPerCluster` automatically. Allow generator to work with empty ranges?

Improve plotFeature.

Add pointsAsNumbers support. Is there a way to facet wrap these instead of using plot grid? Then we can easily support a title. We’re using continuous color here, so the formal won’t match… argument “color” is missing, with no default.

`findMarkers()`: Consider adding `progress` option or BPPARAM support.

Switch to `Misc()` to access the `@misc` slot?

Consider splitting `SeuratMarkers` class into a `DataFrameList`, per cluster.

Stacked bar plot for relative cell abundances per cluster or sample type.

Put the resolution in the plot title for t-SNE.

Add `write = TRUE` argument support for marking looping functions, to write CSVs automatically to disk.

Improve assert checks for `findMarkers()`

syntactic

saveData: Need to harden against accidential pipe to this function.

Check that renaming mode renames symlinks themselves, not the resolved file.

I ran into this attempting to name album artist symlinks in iTunes.

VIGNETTE. Seriously, work on it.

Functions need to convert accent marks if possible. Particularly useful for file names.

Ensure nM returns as nm instead of n_m.

Rename mode: Need to look for and strip ’ - ’ out automatically.

Otherwise this will return ‘-minus-’ in the file name, which is annoying. Need to improve the internal sanitization in R for this.

Rename mode needs to also make extension lowercase.

Need to add recursive rename mode support (koopa only?).

Maybe this is easiest to implement on the R side of things. Need to rename files first, then directories, sorted. Need to work from lowest levels up.

nMito, nCoding looks weird with makeLabel plural.

See bcbioSingleCell example.