NEWS
mappeR 2.0.2
- Really, for real this time, make the default clusterer cut correctly.
mappeR 2.0.1 (2025-02-19)
- Default hierarchical clusterer now correctly cuts dendrograms at the midpoint of their tallest branches.
mappeR 2.0.0 (2025-02-16)
- Clustering is now handled by a
clusterer, which is a function that can handle a list of distance matrices (one for each bin/level set) and output clustering results for each one. The hierarchical clustering included previously is now available as a clusterer called hierarchical_clusterer because I am very creative.
- User-defined clusterers are possible as long as they can handle inputs/outputs correctly. The basic idea is that the output of a
clusterer should look like a list of calls to cutree from the hclust package. Look to the clusterer farm for more examples in the future.
mappeR 1.3.0 (2024-10-30)
- Adjust
compute_tightness to no longer normalize by the maximum distance from the medoid (easier to see behavior in a single mapper graph, may add options in future)
- Add option to consider each level set locally when clustering; default is still to do it globally
mappeR 1.2.0 (2024-10-23)
- Add other hierarchical clustering methods (those available from
fastcluster)
- Fix issues with 100 percent overlap situations caused by
mapply simplifications
mappeR 1.1.0 (2024-10-16)
- Added safety checks for mapper input functions (no
NA inputs, etc)
- Removed Cytoscape interfacing for the release version; will remain in a separate branch
mappeR 1.0.0 (2024-10-07)
- Balls/bins with equal amounts of data no longer cause wonkiness
- Zero overlap situations should work correctly with
igraph
- Added eccentricity lens/filter fuction
mappeR 0.1.5 (2024-10-04)
- 1D tests will not fail due to a one interval cover
mappeR 0.1.4
- Reduced edge widths for Cytoscape visualization.
- Rescaled tightness calculation down a bit.
mappeR 0.1.3 (2024-10-04)
- Edge overlap data is now present in output mapper objects.
- Cytoscape edges are styled by width instead of transparency.
mappeR 0.1.2
- The algorithm now still runs even if there is only one level set to be considered.
mappeR 0.1.1
mappeR 0.1.0