Browse code

Work in progress: refactoring bsseq

- BSseq objects can once again use ordinary matrix objects as assays.
- Reimplement `BSmooth()` more-or-less from scratch:
- Switch from 'parallel' to 'BiocParallel' for parallelization. This brings some notable improvements:
- Smoothed results can now be written directly to an on-disk realization backend by the worker. This dramatically reduces memory usage compared to previous versions of 'bsseq' that required all results be retained in-memory.
- Parallelization is now supported on Windows through the use of a 'SnowParam' object as the value of `BPPARAM`.
- Improved error handling makes it possible to gracefully resume `BSmooth()` jobs that encountered errors by re-doing only the necessary tasks.
- Detailed and extensive job logging facilities.
- Fix bug in `BSmooth()` with the `maxGap` parameter.
- Re-factor BSseq() constructor and add fast, internal .BSseq() constructor.
- Re-factor collapseBSseq() and combine(). Should be much more performant.
- Use beachmat to implement fast validity checking of 'M' and 'Cov' matrices.
- Resave BS.chr22 (supplied data) using integer for storage mode of assays to reduce size.
- Switch from RUnit to testthat. testthat has better integration with code coverage tools that help when refactoring.

Peter Hickey authored on 28/05/2018 23:42:18
Showing 1 changed files
1 1
new file mode 100644
... ...
@@ -0,0 +1,3 @@