from Bio import SeqIO
def calculate_genome_stats(fasta_file):
# Initialize variables to store genome statistics
genome_length = 0
gc_count = 0
a_count = 0
t_count = 0
c_count = 0
g_count = 0
# Read the genome sequence from the FASTA file
for record in SeqIO.parse(fasta_file, "fasta"):
sequence = record.seq
genome_length += len(sequence)
a_count += sequence.count('A')
t_count += sequence.count('T')
c_count += sequence.count('C')
g_count += sequence.count('G')
gc_count += sequence.count('G') + sequence.count('C')
# Calculate GC content
gc_content = (gc_count / genome_length) * 100 if genome_length > 0 else 0
# Print genome statistics
print(f"Genome Length: {genome_length} bp")
print(f"A Count: {a_count}")
print(f"T Count: {t_count}")
print(f"C Count: {c_count}")
print(f"G Count: {g_count}")
print(f"GC Content: {gc_content:.2f}%")
# Example usage
fasta_file = "path/to/your/genome.fasta"
calculate_genome_stats(fasta_file)
