human_129.sort_index(axis=1, inplace=True)
human_129.sort_index(inplace=True)

human_152.sort_index(axis=1, inplace=True)
human_152.sort_index(inplace=True)

intersection_cols = np.intersect1d(human_129.columns, human_152.columns)
intersection_rows = np.intersect1d(human_129.index, human_152.index)

print(len(intersection_rows))
print(len(intersection_cols))

human_129 = human_129.reindex(index=intersection_rows, columns=intersection_cols)
human_152 = human_152.reindex(index=intersection_rows, columns=intersection_cols)

human_129.sort_index(axis=1, inplace=True)
human_129.sort_index(inplace=True)
human_152.sort_index(axis=1, inplace=True)
human_152.sort_index(inplace=True)

print(human_129.shape)
print(human_152.shape)


sparsematrix = io.mmread('my_gene_ontology_edit.obo.2014-01-01.hierarchy_annotations.mtx')
go_hierarchy = sparsematrix.todok()

col_names = row_names = np.genfromtxt('my_gene_ontology_edit.obo.2014-01-01.hierarchy_annotations.mtx.rownames.tsv', dtype=str)

go_hierarchy.axes = col_names

#necessario rinomiare per fare combaciare con termini go
human_129.columns = human_129.columns.str.replace(':', '_')
human_152.columns = human_152.columns.str.replace(':', '_')

#determinare termini go comuni tra organismo e matrice gerarchica
intersection = np.sort(np.intersect1d(go_hierarchy.axes, human_129.columns))
human_129 = human_129.reindex(columns=intersection)
human_152 = human_152.reindex(columns=intersection)

human_129.sort_index(axis=1, inplace=True)
human_129.sort_index(inplace=True)
human_152.sort_index(axis=1, inplace=True)
human_152.sort_index(inplace=True)

#determinazione indici elementi usabili della matrice gerarchica
import numpy_indexed as npi
%time idx = npi.indices(go_hierarchy.axes, intersection)


# seleziona le colonne: attenzione impiega 4-5 minuti con 4 core
%time go__ = go_hierarchy[:,idx]

#seleziona le colonne
%time go = go__[idx,:]

human_129 = csr_matrix(human_129)
human_152 = csr_matrix(human_152)

human_129_tp = (human_129.dot(go) > 0).astype('int8')
human_129_tp = human_129_tp = human_129

human_152_tp = (human_152.dot(go) > 0).astype('int8')
human_152_tp = human_152_tp + human_152

human_129 = pd.DataFrame(human_129_tp.todense(), index=intersection_rows, columns=intersection)
human_152 = pd.DataFrame(human_152_tp.todense(), index=intersection_rows, columns=intersection)