去除比对符号“-”

    
#==========================================
# This code was written by Sihua Peng, PhD.
#==========================================

def clean_sequence(sequence):
    return sequence.replace('-', '')

def clean_fasta(input_filename, output_filename):
    with open(input_filename, 'r') as input_file:
        lines = input_file.readlines()

    cleaned_sequences = []
    current_sequence = ""
    for line in lines:
        if line.startswith('>'):
            if current_sequence:
                cleaned_sequence = clean_sequence(current_sequence)
                cleaned_sequences.append((header, cleaned_sequence))
            header = line.strip()
            current_sequence = ""
        else:
            current_sequence += line.strip()

    if current_sequence:
        cleaned_sequence = clean_sequence(current_sequence)
        cleaned_sequences.append((header, cleaned_sequence))

    with open(output_filename, 'w') as output_file:
        for header, sequence in cleaned_sequences:
            output_file.write(header + '\n')
            output_file.write(sequence + '\n')

input_filename = "L-generated-89-samples.fasta"
output_filename = "L-generated-89-samples-clean.fasta"
clean_fasta(input_filename, output_filename)





    

复制代码