Tutorial: Update kraken databases
15
gravatar for Sej Modha
2.1 years ago by
Sej Modha4.2k
Glasgow, UK
Sej Modha4.2k wrote:

Kraken is a really good k-mer based classification tool. I frequently use this tool for viral signal detection in metagenomic samples. A number of useful scripts such as updating Kraken databases are provided with the tool. Since the NCBI updated the FTP website structure and decided to phase-out Genbank Idenfiers (GIs), the default Kraken database update scripts do not work. So I decided to write a little python script to update Kraken databases that some Kraken users might find useful.

This script automatically downloads:

  • Human genome - most recent assembly version
  • Complete bacterial reference genomes
  • Complete viral reference genomes
  • Archaeal genomes
  • Reference plasmids sequences

To change default genome downloads e.g. download mouse reference genome instead of human; please make necessary changes in the code. Feel free to fork this script on github.

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""

@author: sejalmodha
"""
#import OS module to use os methods and functions
import os
import subprocess
import sys
import re
# you'll see this alias in documentation, examples, etc.
import pandas as pd
#import biopython SeqIO
from Bio import SeqIO

#set present working directory
if len(sys.argv) > 1:
    cwd = sys.argv[1]
else:
    cwd=os.getcwd()
#os.chdir(pwd)

print(cwd)

#get the current working directory 
os.chdir(cwd)
print(os.getcwd())
#function to process ftp url file that is created from assembly files
def process_url_file(inputurlfile):
    url_file=open(inputurlfile,'r')
    file_suffix=r'genomic.gbff.gz'
    for line in url_file:
        url=line.rstrip('\n').split(',')
        ftp_url= url[0]+'/'+url[1]+'_'+url[2]+'_'+file_suffix
        #print(new_url)
        print("Downloading"+ ftp_url)
        #Download the files in the gbff format
        subprocess.call("wget "+ftp_url,shell=True) 
        #unzip the files
        subprocess.call("gunzip *.gz",shell=True)
    return

#function to download bacterial sequences
def download_bacterial_genomes(outfile='outfile.txt'):
    assembly_summary_file=r'ftp://ftp.ncbi.nlm.nih.gov/genomes/refseq/bacteria/assembly_summary.txt'
    if os.path.exists('assembly_summary.txt'):
       os.remove('assembly_summary.txt')
    #Download the file using wget sysyem call
    subprocess.call("wget "+assembly_summary_file, shell=True)
    #Reformat the file to pandas-friendly format
    subprocess.call("sed -i '1d' assembly_summary.txt",shell=True)
    subprocess.call("sed -i 's/^# //' assembly_summary.txt", shell=True)
    #Read the file as a dataframe - using read_table
    #Use read_table if the column separator is tab
    assembly_sum = pd.read_table('assembly_summary.txt',dtype='unicode')
    #filter the dataframe and save the URLs of the complete genomes in a new file
    my_df=assembly_sum[(assembly_sum['version_status'] == 'latest') &
                   (assembly_sum['assembly_level']=='Complete Genome') 
                  ]
    my_df=my_df[['ftp_path','assembly_accession','asm_name']]
    #output_file.write
    my_df.to_csv(outfile,mode='w',index=False,header=None)
    process_url_file(outfile)
    return

#function to download reference genomes 
#this function downloads latest version human reference genome by default 
def download_refseq_genome(taxid=9606,outfile='refseq_genome.txt'):
    assembly_summary_file="ftp://ftp.ncbi.nih.gov/genomes/refseq/assembly_summary_refseq.txt"
    if os.path.exists('assembly_summary_refseq.txt'):
        os.remove('assembly_summary_refseq.txt')
    #Download the file using wget sysyem call
    subprocess.call("wget "+assembly_summary_file, shell=True)
    #Reformat the file to pandas-friendly format
    subprocess.call("sed -i '1d' assembly_summary_refseq.txt",shell=True)
    subprocess.call("sed -i 's/^# //' assembly_summary_refseq.txt", shell=True)
    #Read the file as a dataframe - using read_table
    #Use read_table if the column separator is tab
    assembly_sum = pd.read_table('assembly_summary_refseq.txt',dtype='unicode')
    my_df=assembly_sum[(assembly_sum['taxid'] == taxid) &
                       ((assembly_sum['refseq_category'] == 'reference genome') |
                        (assembly_sum['refseq_category'] == 'representative genome')
                       )]
    my_df=my_df[['ftp_path','assembly_accession','asm_name']]
    #Process the newly created file and download genomes from NCBI website
    my_df.to_csv(outfile,mode='w',index=False,header=None)
    process_url_file(outfile)
    return

#format genbank files to generate kraken-friendly formatted fasta files
def get_fasta_in_kraken_format(outfile_fasta='sequences.fa'):
    output=open(outfile_fasta,'w')
    for file_name in os.listdir(cwd):
        if file_name.endswith('.gbff'):
            records = SeqIO.parse(file_name, "genbank")
            for seq_record in records:
                seq_id=seq_record.id
                seq=seq_record.seq
                for feature in seq_record.features:
                    if 'source' in feature.type:
                        print(feature.qualifiers)
                        taxid=''.join(feature.qualifiers['db_xref'])
                        taxid=re.sub(r'.*taxon:','kraken:taxid|',taxid)
                        print(''.join(taxid))                        
                        outseq=">"+seq_id+"|"+taxid+"\n"+str(seq)+"\n"
                output.write(outseq)
            os.remove(file_name) 
    output.close()  
    return

print('Downloading human genome'+'\n')
#change argument in the following function if you want to download other reference genomes
#taxonomy ID 9606 (human) should be replaced with taxonomy ID of genome of interest
download_refseq_genome(9606,'human_genome_url.txt')
print('Converting sequences to kraken input format'+'\n')
get_fasta_in_kraken_format('human_genome.fa')

#print('Downloading rat genome'+'\n')
#download_refseq_genome(10116,'rat_genome_url.txt')
#print('Converting sequences to kraken input format'+'\n')
#get_fasta_in_kraken_format('rat_genome.fa')

print('Downloading bacterial genomes'+'\n')
download_bacterial_genomes('bacterial_complete_genome_url.txt')
print('Converting sequences to kraken input format'+'\n')
get_fasta_in_kraken_format('bacterial_genomes.fa')

print('Downloading viral genomes'+'\n')
subprocess.call('wget ftp://ftp.ncbi.nlm.nih.gov/refseq/release/viral/viral.*.genomic.gbff.gz',shell=True)
subprocess.call('gunzip *.gz',shell=True)
print('Converting sequences to kraken input format'+'\n')
get_fasta_in_kraken_format('viral_genomes.fa')

print('Downloading archaeal genomes'+'\n')
subprocess.call('wget ftp://ftp.ncbi.nlm.nih.gov/refseq/release/archaea/archaea.*.genomic.gbff.gz',shell=True)
subprocess.call('gunzip *.gz',shell=True)
print('Converting sequences to kraken input format'+'\n')
get_fasta_in_kraken_format('archaeal_genomes.fa')

print('Downloading plasmid sequences'+'\n')
subprocess.call('wget ftp://ftp.ncbi.nlm.nih.gov/refseq/release/plasmid/plasmid.*.genomic.gbff.gz', shell=True)
subprocess.call('gunzip *.gz',shell=True)
print('Converting sequences to kraken input format'+'\n')
get_fasta_in_kraken_format('plasmids_genomes.fa')

#Set name for the krakendb directory
krakendb='HumanVirusBacteria'
subprocess.call('kraken-build --download-taxonomy --db '+krakendb, shell=True)
print('Running Kraken DB build for '+krakendb+'\n')
print('This might take a while '+'\n')
for fasta_file in os.listdir(cwd):
    if fasta_file.endswith('.fa') or fasta_file.endswith('.fasta'):
        print (fasta_file)
        subprocess.call('kraken-build --add-to-library '+fasta_file +' --db '+krakendb,shell=True)
subprocess.call('kraken-build --build --db '+krakendb+' --threads 12',shell=True)
ADD COMMENTlink modified 15 months ago • written 2.1 years ago by Sej Modha4.2k
1

Nice tutorial, i am curious have you tried building a custom database with all virus,bacteria and fungii together?

ADD REPLYlink written 2.0 years ago by badribio240

Hi, I have not built a custom db with fungi but I am sure you can modify this script to do that.

ADD REPLYlink modified 2.0 years ago • written 2.0 years ago by Sej Modha4.2k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 801 users visited in the last hour