def build_cnv_2_gene_training_data(self, data_dir, outcome_file,
cnv_2_gene_file):
'''
Function:
產生以gene symbol為feature的sample set.
Input:
cnv_2_gene_file: 整理好的array_id對應到gene_symbol.
'''
excel_obj = ExcelReader()
data_reader_obj = DataReader()
outcome_dict = excel_obj.get_cyto_cnv_result(outcome_file)
cnv_df = data_reader_obj.cnv_data_reader_pipeline(data_dir)
#### probe mapping to gene
(array_2_gene,
gene_2_array) = data_reader_obj.get_cnv_to_gene_table(cnv_2_gene_file)
# print(array_2_gene)
gene_cnv = data_reader_obj.build_array_to_gene(cnv_df, array_2_gene,
gene_2_array)
## gene cnv
data_df = data_reader_obj.combine_outcome_data(gene_cnv, outcome_dict)
return data_df
def build_cnv_training_data(self, data_dir, outcome_file):
excel_obj = ExcelReader()
data_reader_obj = DataReader()
outcome_dict = excel_obj.get_cyto_cnv_result(outcome_file)
cnv_df = data_reader_obj.cnv_data_reader_pipeline(data_dir)
data_df = data_reader_obj.combine_outcome_data(cnv_df, outcome_dict)
return data_df
def build_probe_2_gene_training_data(self, data_dir, outcome_file,
probe_2_gene_file):
excel_obj = ExcelReader()
data_reader_obj = DataReader()
outcome_dict = excel_obj.get_cyto_cnv_result(outcome_file)
cnv_df = data_reader_obj.cnv_data_reader_pipeline(data_dir)
#### probe mapping to gene
(probe_2_gene, gene_2_probe
) = data_reader_obj.get_probe_to_gene_table(probe_2_gene_file)
gene_cnv = data_reader_obj.build_cnv_to_gene(cnv_df, gene_2_probe)
### probe cnv
# # data_df = data_reader_obj.combine_outcome_data(cnv_df, outcome_dict)
## gene cnv
data_df = data_reader_obj.combine_outcome_data(gene_cnv, outcome_dict)
return data_df
