drcme.ephys_morph_clustering.subsample_run

drcme.ephys_morph_clustering.subsample_run(original_labels, specimen_ids, morph_data, ephys_data, weights=[1, 2, 5], n_cl=[10, 15, 20, 25], n_nn=[4, 7, 10], n_folds=10, n_iter=1, min_consensus_n=3)[source]

Calculate Jaccard coefficients for subsampled clustering runs

Parameters

  • original_labels (array) – Cluster assignments from analysis on full data set

  • specimen_ids (array) – Specimen labels

  • morph_data (array) – Specimen by morphology feature matrix

  • ephys_data (array) – Specimen by electrophysiology feature matrix

  • weights (list, optional) – Set of relative electrophysiology weights

  • n_cl (list, optional) – Set of cluster numbers

  • n_nn (list, optional) – Set of nearest-neighbor values

  • n_folds (int, optional) – Number of subsample folds

  • n_iter (int, optional) – Number of subsampled runs to perform

  • min_consensus_n (int, optional) – Minimum size of consensus cluster

Returns

Jaccard coefficients of each cluster (rows) from each run (columns). This array will have n_iter * n_folds columns.

Return type

array