drcme.bin.run_refine_unstable_coclusters¶

Script to merge cells from unstable clusters into the most similar stable ones.

The script identifies unstable clusters by their Jaccard coefficients falling below a specified threshold (unstable_threshold). It then determines how similar the unstable cluster is to other stable clusters. If it is too dissimilar, it is kept as its own cluster. Otherwise, the unstable cluster is dissolved and its cells are assigned to their best-matching clusters.

It determines whether to dissolve a cluster by calculating whether or not each cell in the unstable cluster has a good match to a stable cluster (i.e., the co-clustering rate exceeds coclust_threshold). If enough of the cells of the unstable cluster have good matches (the fraction of matching cells exceeds pct_needed), the cluster is dissolved and the cells of that cluster are reassigned to stable clusters.

class drcme.bin.run_refine_unstable_coclusters.RefineParameters(extra=None, only=None, exclude=(), prefix='', strict=None, many=False, context=None, load_only=(), dump_only=(), partial=False)[source]¶

Parameter schema for unstable cluster refinement

This schema is designed to be a schema_type for an ArgSchemaParser object

RefineParameters¶
key	description	default	field_type	json_type
input_json	file path of input json file	NA	`InputFile`	str
output_json	file path to output json file	NA	`OutputFile`	str
log_level	set the logging level of the module	ERROR	`LogLevel`	str
cocluster_matrix_file	File path for co-clustering matrix	NA	`InputFile`	str
jaccards_file	File path for Jaccard coefficients	NA	`InputFile`	str
cluster_labels_file	File path for cluster labels	NA	`InputFile`	str
refined_labels_file	Output file path for refined cluster labels (as integers)	NA	`OutputFile`	str
refined_text_labels_file	Output file path for refined cluster labels (as strings with me_prefix)	NA	`OutputFile`	str
refined_ordering_file	Output file path for refined cluster labels	NA	`OutputFile`	str
unstable_threshold	Threshold for Jaccard coefficients to determine stability	0.5	`Float`	float
coclust_threshold	Threshold for co-clustering rate to be considered a match to another cluster	0.4	`Float`	float
pct_needed	Minimum fraction of matching cells to dissolve a cluster	0.33	`Float`	float
me_prefix	prefix for refined cluster text labels	NA	`String`	str

drcme.bin.run_refine_unstable_coclusters.stable_match_rates(clust_labels, shared, stable_clusters)[source]¶

Calculate the co-clustering rates of cells within the stable clusters

Parameters

clust_labels ((n, ) array) – Cluster labels for the n samples
shared ((n, n) array) – Co-clustering rates between all n samples
stable_clusters (list) – List of labels of the stable clusters

Returns

Returns list of the average within-cluster co-clustering rates for every cell found within stable clusters

Return type

list

drcme.bin.run_refine_unstable_coclusters.match_rates_for_unstable_clusters(unstable_clusters, stable_clusters, clust_labels, shared, threshold)[source]¶

Calculate the fraction of cells in unstable clusters that match to a stable cluster.

The highest co-clustering rate with a stable cluster is calculated for each cell in an unstable cluster. If that rate exceeds threshold, that cell is categorized as matching another cluster. The fraction of matching cells is returned for each unstable cluster.

Parameters

unstable_clusters (list) – List of labels of the unstable clusters
stable_clusters (list) – List of labels of the stable clusters
clust_labels ((n, ) array) – Cluster labels for the n samples
shared ((n, n) array) – Co-clustering rates between all n samples
threshold (float) – Minimum co-clustering rate to be considered a match with another cluster

Returns

Dictionary of unstable clusters (keys) and their fractions of matching cells (values)

Return type

dict

drcme.bin.run_refine_unstable_coclusters.new_labels_for_dissolved_cluster(cl, clust_labels, shared, stable_clusters)[source]¶

Relabel the cells in a dissolved cluster with their new assignments

Parameters

cl (int) – Cluster that will be dissolved
clust_labels ((n, ) array) – Cluster labels for the n samples
shared ((n, n) array) – Co-clustering rates between all n samples
stable_clusters (list) – List of labels of the stable clusters

Returns

Array with updated cluster labels

Return type

(n, ) array

Functions

`main`(cocluster_matrix_file, jaccards_file, …)	Main runner function for script.
`match_rates_for_unstable_clusters`(…)	Calculate the fraction of cells in unstable clusters that match to a stable cluster.
`new_labels_for_dissolved_cluster`(cl, …)	Relabel the cells in a dissolved cluster with their new assignments
`stable_match_rates`(clust_labels, shared, …)	Calculate the co-clustering rates of cells within the stable clusters

Classes

RefineParameters([extra, only, exclude, …])

Parameter schema for unstable cluster refinement