An unsupervised kNN Method to systematically detect changes in protein localization in high-throughput microscopy images

Alex X. Lu and Alan M. Moses



  • Python functions for kNN change detection. Requires numpy and scikit learn, as well as
  • Utility functions for unpackaging tab-delimited gene feature files. Requires numpy.
  • Example usage of the above two packages.


  • Profiles.xlsx: Per-gene features (mean, trunucated mean, and median profiling) from single cell data (not included due to size limitations) extracted from the WT3 and RPD3_1 datasets in the CYCLoPs database using image analysis software by Handfield et al. 2015. To use these with the software provided, save individual pages of the spreadsheet as tab-delimited text files.
  • z_scores.xlsx: Processed data (modified z-scores) derived from features in Profiles.xlsx using knn change detection method.