Mercurial > repos > estrain > lissero

annotate variant4b.py @ 0:bc556481a1fb draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
author estrain
date Fri, 13 Mar 2026 12:01:17 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
1 import sys
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
2
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
3 def identify_variants_with_genes(input_file_path, output_file_path, simple_name):
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
4 # Define the genes of interest
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
5 genes_of_interest = ['LMO0737', 'ORF2110', 'ORF2819']
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
6
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
7 # Open the input file and read its lines
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
8 with open(input_file_path, 'r') as file:
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
9 lines = file.readlines()
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
10
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
11 # Check if the file has more than just the header
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
12 if len(lines) <= 1:
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
13 print("Input file does not contain enough data.")
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
14 return
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
15
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
16 # Extract the column headers and find the indices of the genes of interest
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
17 headers = lines[0].strip().split('\t')
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
18 gene_indices = [headers.index(gene) for gene in genes_of_interest]
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
19 serotype_index = headers.index('SEROTYPE')
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
20
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
21 # Modify the header to include the new first column
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
22 modified_header = f"FileName\t{lines[0]}"
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
23
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
24 # Initialize a list to hold the modified lines, starting with the modified header
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
25 modified_lines = [modified_header]
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
26
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
27 # Process each data line in the input file
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
28 for line in lines[1:]:
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
29 data = line.strip().split('\t')
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
30 # Check if the genes of interest are all present (marked as "FULL")
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
31 if all(data[index] == 'FULL' for index in gene_indices):
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
32 # Modify the SEROTYPE column to "4b variant"
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
33 data[serotype_index] = "4b variant"
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
34 # Prepend the simple name to the line
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
35 modified_line = f"{simple_name}\t" + '\t'.join(data) + '\n'
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
36 # Add the modified line to the list
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
37 modified_lines.append(modified_line)
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
38
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
39 # Write the modified lines to the output file
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
40 with open(output_file_path, 'w') as file:
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
41 file.writelines(modified_lines)
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
42
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
43 print(f'Results written to {output_file_path}')
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
44
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
45 if __name__ == "__main__":
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
46 if len(sys.argv) != 4:
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
47 print("Usage: python script.py <input_file_path> <output_file_path> <simple_name>")
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
48 sys.exit(1)
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
49
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
50 input_file_path = sys.argv[1]
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
51 output_file_path = sys.argv[2]
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
52 simple_name = sys.argv[3]
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
53 identify_variants_with_genes(input_file_path, output_file_path, simple_name)
bc556481a1fb planemo upload commit e12eb56d1744da8f7af8ca1819e2617c83fb17a8
estrain
parents:
diff changeset
54