PICASO.kgraph

Classes

`MidpointNormalize`
`KGraph`
`NetworkScorer`	_summary_ Interface class for any network scoring class.
`MeanNetworkScorer`	_summary_ Interface class for any network scoring class.
`NetworkExtender`
`CommunityTool`
`DifferentialCommunityIdentifier`
`GenesetAnnotator`
`DifferentialKG`
`NETSIM`
`ModuleCompare`
`CRankExplorer`
`DefaultDict`	dict() -> new empty dictionary
`TwoLevelDifferentialAnalysis`
`AIDescriptor`

Functions

sjoined(inlist)

Module Contents

class PICASO.kgraph.MidpointNormalize(vmin=None, vmax=None, midpoint=None, clip=False)

Bases: matplotlib.colors.Normalize

midpoint

__call__(value, clip=None)

PICASO.kgraph.sjoined(inlist)

class PICASO.kgraph.KGraph(random_state=42, kgraph_name='KGraph')

kg

random_state

kgraph_name

logger

is_score_field(x)

get_node_attributes()

get_edge_attributes()

__repr__()

copy(suffix=None)

print_kg_info()

load_kgraph_base(data_dir, go=True, TFs=True, omnipath=True, opentargets=True, reactome=True, kegg=True, uniprot_loc=True, STRING=True, NPINTER=False, ot_min_disease_assoc_score=0.8, hallmark_genesets='kegg_gmts/human/c1.all.v2023.2.Hs.symbols.gmt', curated_genesets='kegg_gmts/human/c2.all.v2023.2.Hs.symbols.gmt')

load_kgraph(infile)

save_kgraph(outfile)

get_node_data(node)

get_node_edges(node, src_types=None, tgt_types=None)

get_nodes(nodetype=None)

get_nodes_with_highest_scores(n=3000, nodetype=None, score_accessor=lambda x: ..., node_filter=None)

get_edges_with_highest_scores(n=3000, src_types=None, tgt_types=None, score_accessor=lambda x: ...)

node_types(node, single=False)

get_node_types(single=False)

plot_node_types(show_threshold=0.02)

plot_node_children(node_types=None)

plot_child_distribution(plot=True)

plot_edge_types(field='type', show_threshold=0.02)

plot_edge_between_types(show_threshold=0.02)

_plot_pie(counteritems, show_threshold=0.02)

get_edge_types(field='type')

get_edge_node_types(edge, field='type')

get_edge_edge_types(src_types=None, tgt_types=None, type_accessor='type', ignore_edge_types=None)

get_edge_between_type()

add_gene_expression(exprDF, mean_column='mean', sd_column='sd', perc_expr_column='perc_expr', median_column='median', num_column='num', allnum_column='group_cells')

get_node_type(node)

get_edges_to_type(node, otype)

get_edges_between_ntypes(src_types, tgt_types)

_get_predecessors(start_node, ntype, n=10)

score_nodes_hierarchically(ntype='geneset', target_ntype='gene', relevance_threshold=0, child_score_accessor=lambda x: ...)

node_type_overlap(node, types)

get_node_scores(score_accessor=lambda x: ..., nodes=None)

get_edge_scores(score_accessor=lambda x: ..., edge_types=None, nodes=None)

get_edge_scores_per_type(score_accessor=lambda x: ..., edge_types=None, single=False)

plot_score_histogram(edge_types=None, score_accessor=lambda x: ...)

plot_edge_attribute_distribution(edge_types=None, score_accessor=lambda x: ..., ax=None, title=None)

plot_node_attribute_distribution(attribute_accessor, node_types=None, ax=None, title=None)

plot_node_attribute_histogram(attribute_accessor, nodes=None, node_type=None, ax=None, title=None)

plot_score_violin(per_edge_type=False, single_edge_types=False, edge_types=None, score_accessor=lambda x: ..., figsize=None)

plot_subgraph_scores(scores)

filter_nodes(filter_function)

filter_edges(filter_function)

subset_kg(retainedNodes, suffix='subset')

induce_kg(retainedNodes, suffix='subset', radius=1)

to_gene_kgraph()

_filter_edge_score(score_field, minEdgeScore=3.0, undirected=True)

get_communities_connectedcomponent(minEdgeScore=3.0, resolution=5, prefix='Community', sep='_', score_field='score')

get_communities_greedymodularity(minEdgeScore=3.0, resolution=0.5, prefix='Community', sep='_', score_field='score')

get_communities_asyn_label_propagation(minEdgeScore=3.0, prefix='Community', sep='_', score_field='score', seed=None)

get_communities(minEdgeScore=3.0, resolution=5, prefix='Community', sep='_', score_field='score')

get_communities_ecg(minEdgeScore=3.0, resolution=5, ens_size=16, prefix='Community', sep='_', score_field='score')

get_communities_negpos(max_comm_size=50, prefix='Community', sep='_', score_field='score')

get_communities_infomap(prefix='Community', sep=','): Partition network with the Infomap algorithm. Annotates nodes with ‘community’ id and return number of communities found.

get_communities_markovclustering(prefix='cluster', inflation=1.4)

get_communities_link(minEdgeScore=3.0, threshold=0.15, score_field='score', prefix='Community', sep='_')

describe_communities(comms)

get_kg_subgraph(genes)

plot_graph_network(outfile=None, notebook=False)

get_nx_subgraph(genes)

plot_graph(ax=None, figsize=(6, 6), title='', pos=None, close=True, font_size=8, edge_score_normalizer=None, node_score_normalizer=None, edge_cmap=plt.cm.Reds, max_node_size=200, nodetype2color={'gene': '#239756', 'geneset': '#3fc37e', 'disease': '#5047ee', 'drug': '#3026c1', 'NA': '#f37855'}, nodecolors={'gene': '#239756', 'geneset': '#3fc37e', 'disease': '#5047ee', 'drug': '#e600e6', 'NA': '#f37855'}, nodeshapes={'gene': 'o', 'geneset': 's', 'disease': '^', 'drug': 'p', 'NA': 'o'}, edge_score_accessor=lambda x: ..., node_score_accessor=lambda x: ..., legend=True)

get_node_degrees(intype, connecttype=None)

class PICASO.kgraph.NetworkScorer(random_state=42)

_summary_ Interface class for any network scoring class.

Hint: All scores must have their minimum at 0 ! This is required such that differential KGs can be successfully generated!

random_state

logger

score()

class PICASO.kgraph.MeanNetworkScorer(random_state=42)

Bases: NetworkScorer

_summary_ Interface class for any network scoring class.

Hint: All scores must have their minimum at 0 ! This is required such that differential KGs can be successfully generated!

scoring_gene_gene_expression

scoring_interactions

score_edges(kgraph, value_accessor, edge_accessor, scorers, src_types=None, tgt_types=None, ignore_edge_types=None)

score_nodes_from_properties(kgraph, ntypes=['measured_expression'], scoring_function=lambda x: ..., property_accessor=lambda x: ...)

score_nodes_from_edges(kgraph, ntype='geneset', consider_edge_type=[('gene', 'geneset')], scoring_function=None, overwrite_score=False)

score_nx(kg: networkx.DiGraph)

score(kgraph: KGraph)

calculate_edge_zscores(kgraph, score_accessor, edge_accessor, src_types=None, tgt_types=None, ignore_edge_types=None)

class PICASO.kgraph.NetworkExtender

extend_network(nodes, fullKG: KGraph, radius=1, scorer: NetworkScorer = None, min_children_gs=2, max_size_gs=100, minFraction_small=0.4, minFraction_large=0.5, node_types=['geneset', 'disease', 'ncRNA'], minGeneSpec={'geneset': 0.8, 'disease': 0.6}, min_edge_score=1.0, score_field='score', verbose=False)

extend_network_force(eKG: KGraph, fullKG: KGraph, nodetype, acceptor=None, edge_acceptor=None, minSpec=0.5)

extend_nodetypes(eKG: KGraph, fullKG: KGraph, nodetype, node_score_accessor=None, edge_score_accessor=None, min_node_score=0.5, verbose=False)

class PICASO.kgraph.CommunityTool

scatter_community_scores(module_detail)

visualize_communities(details, title, subsetOrderFunc=None, field='median', show_values=True)

sort_communities(comm_details, details=False)

plot_community(KGs, communityNodes, own, main_net=None, num_columns=4, title=None, nodecolors={'gene': '#239756', 'geneset': '#3fc37e', 'disease': '#5047ee', 'drug': '#3026c1', 'NA': '#f37855'}, outfile=None, dpi=500, show=True, edge_score_accessor=lambda x: ..., node_score_accessor=lambda x: ..., verbose=False)

compare_modules(comms, figsize=(16, 12))

class PICASO.kgraph.DifferentialCommunityIdentifier

classmethod cohend(d1, d2)

calculate_scores(allKGs, ref_kg, nodes, score_field)

identify_differential_communities(communities, ref_kg, KGs, sort_function=None, score_field='score', use_statistic='cohend', min_nodes=10, min_enriched=0.5, min_effect_size=0.2, all_verbose=False, verbose=False)

score_subgraphs_for_subnet(KGs, subnet, score_field='score')

score_subgraphs(Gs, score_field='score', default_score=0)

class PICASO.kgraph.GenesetAnnotator

annotate_genesets(kg: KGraph, settype='disease', targettype='gene')

class PICASO.kgraph.DifferentialKG(pseudocount=0.01)

pseudocount

_get_edge_fold_changes(kg1: KGraph, kg2: KGraph)

_get_log_foldchange(node1Score, node2Score, log2=True)

_get_node_fold_changes(kg1: KGraph, kg2: KGraph)

get_differential_graph(kg1: KGraph, kg2: KGraph, field='fc', rescoring=['geneset', 'disease'])

calculate_diffkg_list(exprKGs, base_case, cases)

class PICASO.kgraph.NETSIM(kg: KGraph, conf_score_accessor=lambda x: ...)

kg

orig_kg

conf_score_accessor

kg_rev

undirected_x

max_go_level = 10

precalculated_term_genes

get_annotated_genes(ta)

confidence(gi, gj)

dij(gi, gj)

functional_distance_genesets(ta, tb, ga, gb)

get_lca(ta, tb)

get_shortest_path(ta, tb)

get_path_constrained_annotation(ta, tb, p, verbose=False)

get_node_name(node)

functional_similarity(ta, tb, verbose=False)

get_go_level(ta)

lca_similarity(ta, tb, verbose=False)

get_relevant_goterms(groupKG, max_terms=None, verbose=False)

compare_modules_lca(skg1, skg2, max_terms=None, verbose=True)

compare_modules(skg1, skg2, max_terms=None, verbose=True)

class PICASO.kgraph.ModuleCompare

classmethod makeProgressBar() → progressbar.ProgressBar

draw_network(G, title=None, borderWeightQuantile=0.8)

module_similarities_to_df(modSims)

network_compare_modules(inKGs, measure='jaccard', borderWeightQuantile=0.8)

network_compare_lca(inKGs, max_terms=None, fullKG=None, ns=None, borderWeightQuantile=0.8)

network_compare_netsim(inKGs, borderWeightQuantile=0.8, max_terms=None, fullKG=None, ns=None)

module_pca(sigKG, kg: KGraph)

plot_dendrogram(simDict, figsize=(8, 6), color_threshold=1)

class PICASO.kgraph.CRankExplorer

calculate_connectivity(kg: KGraph)

calculate_density(communityKG: KGraph, edge_score_accessor)

calculate_boundary(community, fullKG: KGraph, edge_score_accessor, density)

calculate_allegiance(community, fullKG: KGraph)

evaluate_community(community, fullKG: KGraph, edge_score_accessor=lambda x: ...)

evaluate_communities(sigKGs, fullKGs, mod_sep='_mod_')

plot_scores(scoreDF)

class PICASO.kgraph.DefaultDict

Bases: dict

dict() -> new empty dictionary dict(mapping) -> new dictionary initialized from a mapping object’s

(key, value) pairs

dict(iterable) -> new dictionary initialized as if via:: d = {} for k, v in iterable:

d[k] = v
dict(**kwargs) -> new dictionary initialized with the name=value pairs: in the keyword argument list. For example: dict(one=1, two=2)

__missing__(key)

class PICASO.kgraph.TwoLevelDifferentialAnalysis(tlDict, sorted_zones, output_folder_formatter, fullKG=None)

tldict

sorted_zones

name_sep = '_mod_'

cellgroupdata

output_folder_formatter

recalc_warning = False

calculate_modules(relevant_cellgroups=None, cg_zone_formatter='{}_{}', reference_formatter='{}_Ref', diffkg=DifferentialKG(), overwrite=False)

_get_diff_comms(tkgs, min_effect_size=1.0, resolution=4, minEdgeScore=1, min_node_scores={'drug': 1.0, 'ncRNA': 0.7}, network_extend_spec={'geneset': {'min_gene_spec': 0.5, 'max_size_gs': 200, 'min_fraction_large': 0.6, 'min_fraction_small': 0.5}, 'disease': {'min_gene_spec': 0.8, 'max_size_gs': 100, 'min_fraction_large': 0.7, 'min_fraction_small': 0.6}})

plot_module_comparisons(plot_communities=False, ct=CommunityTool())

_geneset_filter(n, kg)

_filter_empty_genesets(n, kg)

_filter_singletons(n, kg)

_combined_filter(n, kg)

property communities

get_community(cid)

get_community_details(cid)

get_community_kg(cid)

get_community_representation(cid, key='communities')

plot_module_description(nrow=4, figsize=(10, 10), dcolors=None)

_describe_kg(kg, name)

describe_modules(relevant_cellgroups=None, non_verbose=False)

create_overlap_df(nodetype)

_calculate_gs_overlap(gsType, gsDict, moduleKG)

_toShortName(x)

describe_module_scrna(adata, sc_celltype_column, sc_condition_column, show_plot=True, plot_folder=None, plot_prefix='overview_plot_{}', module_names=None, gsTypes=['disease', 'geneset', 'drug'], numGenesThreshold=3, numElemsBarPlot=10, hue_colors={'gene': '#239756', 'geneset': '#3fc37e', 'disease': '#5047ee', 'drug': '#e600e6', 'NA': '#f37855'})

available_cellgroups()

class PICASO.kgraph.AIDescriptor(model_name='LoneStriker/BioMistral-7B-DARE-GGUF', model_file='BioMistral-7B-DARE-Q4_K_M.gguf', local_dir='../llm_addon/model/')

model_name

model_file

local_dir

model_path

llm

query_genelist(gene_list, context=None, verbose=False, word_cloud=False)

_plot_wordcloud(text)

query_wordcloud(gene_list, context)