------------------
* Added sumo icon.
* Implemented new mode of sumo *interpret*, for finding features that drive clusters separation.
* Added example test case and data preprocessing suggestions in documentation.
[*prepare*]
* Improved plotting of heatmaps.
* Changed acceptable data types to binary [0, 1] data and continuous data which is standardized feature wise.
* Added cosine similarity measure, recommended for sparse data.
* Removed variable type argument, categorical and binary distance measure. Similarity measures now can be set for every layer separately with *-method* parameter. Available measures include pearson and spearman correlation, (new) cosine similarity and euclidean distance with RBF kernel (before used for continuous variables). The last measure remain the default.
* Updated support for .txt (space) and .tsv (tab delimited files).
* Added support for compressed files (.txt.gz, .txt.bz2, .tsv.gz, .tsv.bz2).
* Removed support for .npz files.
[*run*]
* Fixed typo resulting in adding two identical consensus matrices into .npz file.
* Changed default sparsity from a range of values to one (0.1), as *sumo* results are very stable towards changes in sparsity parameter (unless the number of clusters is raised to very high values).
* Improved plotting of consensus matrices.
* Added proportion of ambiguous clusterings (PAC) plot for improved number of clusters selection.
* Changed sumo_results.npz file into symbolic link to selected eta result file.
* Added text file with cluster labels in every k directory.
[*evaluate*]
* Change mode to support .tsv files as both input and labels file.