Fixed a bug in v1.10.0 related to generating the plots PDF file.
Now Clust can run on Python 3 as well as Python 2!
Update: a bug was found in this version related to creating the plots PDF file. Fixed in v1.10.1.
Enables Python users to run clust within scripts as the "runclust" function which accepts dataframes or arrays as an inputs and gives dataframes as outputs. Previously, clust was only runnable from terminals with input/output files.
Users running clust from terminal will not be affected (except for minor bug fixes and email updates).
Fixed a bug when some datasets are not represented in the replicates file
Fixed a bug when data comes from multiple species with missing genes from some species
Cluster visualisation enhanced and some other minor fixes done.
Fixed a bug that happens in the rare case when clust generates zero clusters or when the entire input dataset is filtered out before clustering.
If a single dataset is analysed, the user now can give the dataset's file's path directly to the clust command without having to include it in a data directory
Updated auto-normalisation for two-sided data
Changed the citation from the bioRxiv preprint to the published Genome Biology paper
- Fixed bugs when data files have UNICODE characters
- Results Summary.tsv file includes more metrics (e.g. number of conditions and samples)
- Results clusters profiles PDF file is more tidy with y-ticks included
Faster clust! Default parameters edited to make it much faster without compromising quality.
Feature added: Automatic normalisation!
Feature added: Automatic filtering out of genes with flat profiles!
Allowed for values in data files to be delimited with tabs, spaces, commas, semicolons, or mixtures of those.
Updated the reference to the bioRxiv new preprint.
Added the bioRxiv preprint citation.
Partially fixed zero clusters issue & moved quantile norm. before summarisation and filtering
Enhanced cluster separability in the optimisation step
Added the deterministic option and coded the deterministic Kaufman's initialisation algorithm for k-means, in a memory-friendly manner.
Resolved a memory error caused by hierarchical clustering over large datasets
Further bugs were resolved!
Few more bugs have been fixed especially while analysing multiple datasets from different species.