cosifer.collections package

Submodules

cosifer.collections.graph module

Graph class.

class cosifer.collections.graph.Graph(adjacency, labels=None, undirected=True)

Bases: object

Lightweight class for handling cosifer graph objects. If the graph is undirected the adjacency matrix is stored as a sparse lower triangular matrix.

n

number of nodes.

Type

int

labels_to_indices

label to index mapping.

Type

pd.Series

indices_to_labels

index to lable mapping.

Type

pd.Series

adjacency

sparse adjacency.

Type

spicy.sparse.csr_matrix

undirected

flag indicating whether edges are directed.

Type

bool

get_scaled_adjacency()

Get a min-max scaled version of the adjacency.

Returns

the min-max scaled adjacency.

Return type

spicy.sparse.csr_matrix

indices_to_labels = None

labels_to_indices = None

n = 0

set_adjacency_from_numpy(adjacency)

Set the adjacency from a numpy ndarray.

Parameters

adjacency (np.ndarray) – adjacency matrix.

Raises

RuntimeError – in case of inconsitencies in the sizes.

set_adjacency_from_pandas(adjacency)

Set the adjacency from a pandas dataframe.

Parameters

adjacency (pd.DataFrame) – adjacency matrix.

Raises

RuntimeError – in case of inconsitencies in the sizes.

set_adjacency_from_sparse(adjacency)

Set the adjacency from a sparse matrix.

Parameters

adjacency (spicy.sparse.csr_matrix) – adjacency matrix.

Raises

RuntimeError – in case of inconsitencies in the sizes.

set_labels(labels)

Set the node labels.

Parameters

labels (iterable) – labels to set.

Raises

RuntimeError – in case of inconsitenties between the labels and the graph nodes.

to_interaction_table(scaled=True, interaction_symbol='<->')

Convert the graph to an interaction table.

Parameters

• scaled (bool, optional) – flag to activate min-max scaling of the edges. Defaults to True.

• interaction_symbol (str, optional) – symbol to depict interactions between labels. Defaults to ‘<->’.

Returns

the table containing the interactions reported in

the graph.

Return type

InteractionTable

cosifer.collections.interaction_table module

Interation table class.

class cosifer.collections.interaction_table.InteractionTable(df=Empty DataFrame Columns: [e1, e2, intensity] Index: [], labels=None, interaction_symbol=None, force_undirected=False)

Bases: object

Interaction table class representation of an edge list.

df

underlying dataframe containing the edge list.

Type

pd.DataFrame

labels

node labels.

Type

iterable

apply_filter(labels=None, prune_labels=True, indices=None, threshold=0.0, top_n=None)

Apply a filter to get an InteractionTable.

Parameters

• labels (iterable, optional) – node labels to select. Defaults to None.

• prune_labels (bool, optional) – prune labels not present in the final table. Defaults to True.

• indices (itarable, optional) – indices to filter rows. Defaults to None.

• threshold (float, optional) – threshold for the edge weights. Defaults to 0.0.

• top_n (int, optional) – number of top interactions to keep. Defaults to None.

Returns

a filtered interaction table.

Return type

InteractionTable

get_df_dict(threshold=None)

Get a dictionary to represent the edge list.

Parameters

threshold (float, optional) – threshold to filter edge by intensity. Defaults to None.

Returns

a dictionary representing then edge list dataframe.

Return type

dict

to_edge_list(interaction_symbol='<->', weights=True)

Returns an edge list containing the interactions.

Parameters

• interaction_symbol (str, optional) – Symbol separating the labels in the index of the edge list dataframe. Defaults to ‘<->’.

• weights (bool, optional) – Flag indicating whether weights are returned. Defaults to True.

Returns

a list of edges represented by tuples.

Return type

list

to_graph(undirected=True, imposed_labels=None)

Get a graph from the stored edges.

Parameters

• undirected (bool, optional) – flag to indicate whether the interactions are undirected. Defaults to True.

• imposed_labels (iterable, optional) – node label to consider. Defaults to None.

Returns

a graph.

Return type

Graph

cosifer.collections.interaction_table.directed_to_undirected_interactions(directed_interactions)

Processing of a directed table, discarding directions and keeping only the maximum edge value. It will drop and reindex with integers the table. :param: directed_interactions: directed interactions table. :return: an undirected version of the given table.

Parameters

directed_interactions (pd.DataFrame) – directed interactions dataframe.

Returns

undirected interactions dataframe

Return type

pd.DataFrame

cosifer.collections.interaction_table.interaction_table_from_dict(interaction_dictionary)

Construct an InteractionTable from a dictionary.

Parameters

interaction_dictionary (dict) – an interaction dictionary. The dictionary should contain two keys: ‘labels’, containing the node labels, and ‘interactions’, an object that can be used to construct a dataframe representing the edge list.

Returns

the interaction table.

Return type

InteractionTable

cosifer.collections.interaction_table.interaction_table_from_edge_list(interaction_list)

Construct an InteractionTable from a list.

Parameters

interaction_list (list) – an edge list containing tuples.

Returns

the interaction table for the provided edge list.

Return type

InteractionTable

cosifer.collections.interaction_table.interaction_table_from_gzip(filepath)

Construct an InteractionTable from a gzipped file containing an edge list.

Parameters

filepath (str) – path to the gipped file.

Returns

the interaction table from the gzipped edge list.

Return type

InteractionTable

cosifer.collections.interaction_table.interaction_table_to_edge_list(interaction_table, interaction_symbol='<->', weights=True)

Convert an InteractionTable to an edge list.

Parameters

• interaction_table (InteractionTable) – an interaction table.

• interaction_symbol (str, optional) – Symbol separating the labels in the index of the edge list dataframe. Defaults to ‘<->’.

• weights (bool, optional) – Flag indicating whether weights are returned. Defaults to True.

Returns

a list of edges represented by tuples.

Return type

list

cosifer.collections.interaction_table.process_group(row)

Process a grouped edge list dataframe

Parameters

row (pd.Series) – a dataframe row.

Returns

a list with entitity labels and weight.

Return type

list

cosifer.collections.interaction_table.process_index(index, intensity, interaction_symbol)

Process index to get edge tuple. Disregard edge weight.

Parameters

• index (str) – index of the edge list dataframe.

• intensity (float) – intensity of the edge.

• interaction_symbol (str) – symbol used to separate node labels.

Returns

a tuple containing edge labels.

Return type

tuple

cosifer.collections.interaction_table.process_index_with_weights(index, intensity, interaction_symbol)

Process index to get edge tuple including edge weight.

Parameters

• index (str) – index of the edge list dataframe.

• intensity (float) – intensity of the edge.

• interaction_symbol (str) – symbol used to separate node labels.

Returns

a tuple containing edge labels and weight.

Return type

tuple

Module contents