Clustering — mdpath.src.cluster

This module contains the class PatwayClustering which calculates the overlap between pathways and clusters them based on the overlap. Clusters are generated through hirarcical clustering using scipy. Optimal cluster size is evaluated using the silhouette score.

Classes

PatwayClustering

class mdpath.src.cluster.PatwayClustering(df_close_res: DataFrame, pathways: list, num_processes: int)

Bases: object

Perform clustering of pathways based on the overlap of close residue pairs.

Attributes:

df (pd.DataFrame): DataFrame containing close residue pairs.

pathways (list): List of pathways, where each pathway is a list of residue indices.

num_processes (int): Number of processes to use for parallel computation.

overlapp_df (pd.DataFrame): DataFrame containing the overlap between all pathway pairs.

calculate_overlap_for_pathway(args: tuple) → list

Calculates the overlap between a pathway and all other pathways.

Args:

args (tuple): Argument wrapper conatining the pathway index, the pathway, all pathways and the dataframe with close residue pairs.

Returns:

result (list): List of dictionaries with the overlap between the given pathway and all other pathways.

calculate_overlap_parallel() → DataFrame

Parallelization wrapper for the calculate_overlap_for_pathway function.

Returns:

overlap_df (pd.DataFrame): Pandas dataframe with the overlap between all pathways and all other pathways.

pathway_clusters_dictionary(clusters: dict, sorted_paths: list) → dict

Generates a dictionary mapping cluster numbers to lists of pathways.

Args:

clusters (dict): A dictionary where keys are cluster numbers and values are lists of pathway IDs.

sorted_paths (list): A list of pathways, where each pathway is a tuple and the first element is the pathway name.

Returns:

dict: A dictionary where keys are cluster numbers and values are lists of pathways corresponding to each cluster.

pathways_cluster(n_top_clust: int = 0, save_path: str = 'clustered_paths.png') → dict

Clustering of pathways based on the overlap between them.

Args:

n_top_clust (int, optional): Number of clusters to output. Defaults to all.

save_path (str, optional): Save path for cluster dendogram figure. Defaults to “clustered_paths.png”.

Returns:

clusters (dict): Dictionary with the clusters and their pathways.