import sys
import os

#path = "/Users/rmadupuri/Desktop/case_lists_test"
path = "/Users/rmadupuri/GitHub/datahub/public"

#dictionary of case_list and the corresponding name
case_names = {
	"cases_all.txt": "All samples",
	"cases_sequenced.txt": "Samples with mutation data",
	"cases_cna.txt": "Samples with CNA data",
	"cases_cnaseq.txt": "Samples with mutation and CNA data",
	"cases_log2CNA.txt": "Samples with log2 copy-number data",
	"cases_methylation_all.txt": "Samples with methylation data",
	"cases_methylation_hm27.txt": "Samples with methylation data (HM27)",
	"cases_methylation_hm450.txt": "Samples with methylation data (HM450)",
	"cases_miRNA.txt": "Samples with microRNA data (microRNA-Seq)",
	"cases_mRNA_U133.txt": "Samples with mRNA data (U133 microarray)",
	"cases_mRNA.txt": "Samples with mRNA data (Agilent microarray)",
	"cases_protein_quantification.txt": "Samples with protein data (Mass Spec)",
	"cases_RNA_Seq_mRNA.txt": "Samples with mRNA data (RNA Seq)",
	"cases_RNA_Seq_v2_mRNA.txt": "Samples with mRNA data (RNA Seq V2)",
	"cases_rppa.txt": "Samples with protein data (RPPA)",
	"cases_complete.txt": "Complete samples"
  }

#dictionary of case_list and their corresponding descriptions
case_desc = {
	"cases_all.txt": "All samples",
	"cases_sequenced.txt": "Samples with mutation data",
	"cases_cna.txt": "Samples with CNA data",
	"cases_cnaseq.txt": "Samples with mutation and CNA data",
	"cases_rppa.txt": "Samples protein data (RPPA)",
	"cases_log2CNA.txt": "Samples with log2 copy-number data",
	"cases_methylation_all.txt": "Samples with methylation data",
	"cases_methylation_hm27.txt": "Samples with methylation data (HM27)",
	"cases_methylation_hm450.txt": "Samples with methylation data (HM450)",
	"cases_miRNA.txt": "Samples with microRNA data",
	"cases_mRNA_U133.txt": "Samples with mRNA expression data",
	"cases_mRNA.txt": "Samples with mRNA expression data",
	"cases_protein_quantification.txt": "Samples with protein data (Mass Spec)",
	"cases_RNA_Seq_mRNA.txt": "Samples with mRNA expression data",
	"cases_RNA_Seq_v2_mRNA.txt": "Samples with mRNA expression data",
	"cases_complete.txt": "Samples with mutation, CNA and expression data"
  }


all_studies = os.listdir(path)

for st in all_studies:
	if st != ".DS_Store":
		files = os.listdir(path+"/"+st)
		if "case_lists" in files:
			case_list_files = os.listdir(path+"/"+st+"/case_lists")
			for val in case_names:
				if val in case_list_files:
					seq_data = str()
					with open(path+"/"+st+"/case_lists"+"/"+val,'r') as seqf:
						for line in seqf:
							if line.startswith("case_list_name:"):
								seq_data += "case_list_name: "+ case_names[val] +"\n"
							elif line.startswith("case_list_description:"):
								description = line.rstrip('\n').split(' ')
								seq_data += "case_list_description: "+case_desc[val]+" "+description[len(description)-2]+" "+description[len(description)-1]+"\n"
							else:
								seq_data += line
					os.remove(path+"/"+st+"/case_lists"+"/"+val)
					f= open(path+"/"+st+"/case_lists"+"/"+val,"w+")
					f.write(seq_data)