angelolab / ark-analysis / 7329

93%

master: 94%

push

travis-ci-com

Add native Python consensus clustering process (#839)

* Add initial implementation of consensus clustering class

* Add initial implementation of pixel consensus clustering in

* Add initial implementation of cell consensus clustering in

* Change call to scale_data

* Remove inplace argument from zscore apply

* Add status updates to consensus clustering

* Fix call to verify_in_list

* Add some more comments to PixieConsensusCluster

* Documentation, PYCODESTYLE, and add cluster_type as parameter to the constructor of PixieConsensusCluster

* Bring out som_col and meta_col definitions to the constructor

* More documentation

* Ensure cluster variable is type hinted to Callable

* Patch up feather save call

* Documentation OCD

* Make sure fovs_list instead of fov_list is used

* Fix saving of consensus mapping to .feather

* Multiprocessing fixes

* Wrap consensus cluster call around helper function

* Adjust mapping scheme

* Ensure meta cluster values are corrected to 1-indexing

* Progress outputs

* Make sure progress updates for multiprocessing print out every batch and not just final

* Set random seed for consensus clustering

* Port over changes in pixel clustering to cell clustering

* Remove unnecessary ncores parameter

* Use new multiprocess logic

* Enforce consistent pixel_data_path naming

* Typo: fov_list to fovs_list

* Remove pandas copy warning from meta cluster adding 1 correction

* Make sure cell data is being loaded in as feather and not csv

* Make sure cell_cluster_utils imports cluster_helpers

* Patch up input file to cell consensus clustering

* Need to specify cell consensus cluster column filtering as regex

* Add initial implementation for testing PixieConsensusCluster

* Nuke consensus clustering R files

* Remove old error messages from multiprocessing pixel consensus clustering that will no longer appear with nuke of R

* Run both pixel and c... (continued)

162 of 162 new or added lines in 3 files covered. (100.0%)

3310 of 3530 relevant lines covered (93.77%)

0.94 hits per line