comparison COBRAxy/ras_generator.py @ 513:b02cfa3b36dd draft default tip

Uploaded

512:f32d3c9089fc

    rules_total_string=list(set(rules_total_string.split(" ")))

    if any(dataset.index.duplicated(keep=False)):
        genes_duplicates=orig_gene_list[dataset.index.duplicated(keep=False)]
        genes_duplicates_in_model=[elem for elem in genes_duplicates if elem in rules_total_string]
        if len(genes_duplicates_in_model)>0:#metabolic genes have duplicated entries in the dataset
            list_str=", ".join(genes_duplicates_in_model)
            raise ValueError(f"ERROR: Duplicate entries in the gene dataset present in one or more GPR. The following metabolic genes are duplicated: "+list_str)       

    #check if nan value must be ignored in the GPR 
    if ARGS.none:
    #    #e.g. (A or nan --> A)
        ignore_nan = True

    #report genes not present in the data
    if len(genes_not_mapped)>0: 
        genes_not_mapped_str=", ".join(genes_not_mapped)
        utils.logWarning(
        f"The following genes are mentioned in the GPR rules but don't appear in the dataset: "+genes_not_mapped_str,
        ARGS.out_log)  
 
    print("Execution succeeded")

###############################################################################

513:b02cfa3b36dd

    rules_total_string=list(set(rules_total_string.split(" ")))

    if any(dataset.index.duplicated(keep=False)):
        genes_duplicates=orig_gene_list[dataset.index.duplicated(keep=False)]
        genes_duplicates_in_model=[elem for elem in genes_duplicates if elem in rules_total_string]

        if len(genes_duplicates_in_model)>0:#metabolic genes have duplicated entries in the dataset
            list_str=", ".join(genes_duplicates_in_model)
            list_genes=f"ERROR: Duplicate entries in the gene dataset present in one or more GPR. The following metabolic genes are duplicated: "+list_str
            raise ValueError(list_genes)       
        else:
            list_str=", ".join(genes_duplicates)
            list_genes=f"INFO: Duplicate entries in the gene dataset. The following genes are duplicated in the dataset but not mentioned in the GPRs: "+list_str           
            utils.logWarning(list_genes,ARGS.out_log)  

    #check if nan value must be ignored in the GPR 
    if ARGS.none:
    #    #e.g. (A or nan --> A)
        ignore_nan = True

    #report genes not present in the data
    if len(genes_not_mapped)>0: 
        genes_not_mapped_str=", ".join(genes_not_mapped)
        utils.logWarning(
        f"INFO: The following genes are mentioned in the GPR rules but don't appear in the dataset: "+genes_not_mapped_str,
        ARGS.out_log)  
 
    print("Execution succeeded")

###############################################################################