##Supplemental Data 6: The bash script used in extracting the coding sequences from the genome.

from Bio import SeqIO
from Bio.SeqRecord import SeqRecord
from Bio.Seq import Seq

# Function to remove stop codons from a nucleotide sequence
def remove_stop_codons(sequence):
    stop_codons = ["TAA", "TAG", "TGA"]
    for codon in stop_codons:
        sequence = sequence.replace(codon, "")
    return sequence

# Function to extract unique protein-coding genes and save them to all_genes.fasta
def extract_and_save_protein_coding_genes(genbank_file, output_fasta):
    unique_genes = set()
    gene_records = []

    for record in SeqIO.parse(genbank_file, "genbank"):
        for feature in record.features:
            if feature.type == "CDS":
                if "gene" in feature.qualifiers:
                    gene_name = feature.qualifiers["gene"][0]
                    if gene_name not in unique_genes:
                        unique_genes.add(gene_name)
                        gene_seq = feature.location.extract(record).seq
                        gene_record = SeqRecord(gene_seq, id=gene_name, description="")
                        gene_records.append(gene_record)

    SeqIO.write(gene_records, output_fasta, "fasta")

    return gene_records

# Provide the path to your GenBank file
genbank_file = "14.gb"

# Output file for all extracted genes
all_genes_output_fasta = "all_genes.fasta"

# Extract and save protein-coding genes to all_genes.fasta
gene_records = extract_and_save_protein_coding_genes(genbank_file, all_genes_output_fasta)

# Concatenate gene sequences, remove stop codons, and save as concatenated.fasta
concatenated_sequence = ""
for record in gene_records:
    concatenated_sequence += str(record.seq)

# Remove stop codons from the concatenated sequence
concatenated_sequence_no_stop = remove_stop_codons(concatenated_sequence)

# Write the concatenated sequence without stop codons to a file
concatenated_output_fasta = "concatenated.fasta"
with open(concatenated_output_fasta, "w") as output_handle:
    output_handle.write(">Concatenated_Protein_Coding_Sequences\n")
    output_handle.write(concatenated_sequence_no_stop)

print("All extracted protein-coding genes have been written to", all_genes_output_fasta)
print("Concatenated nucleotide sequences of protein-coding genes with stop codons removed have been written to", concatenated_output_fasta)