MetaMeta: Integrating metagenome analysis tools to improve taxonomic profiling

Vitor C. Piro ([email protected])

Piro, V. C., Matschkowski, M., & Renard, B. Y. (2017). MetaMeta: integrating metagenome analysis tools to improve taxonomic profiling. Microbiome, 5(1), 101. http://doi.org/10.1186/s40168-017-0318-y

Install:

Miniconda:

# Download conda installer
wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh

# Set permissions to execute
chmod +x Miniconda3-latest-Linux-x86_64.sh 	

# Execute. Make sure to "yes" to add the conda to your PATH
./Miniconda3-latest-Linux-x86_64.sh 		

# Add channels
conda config --add channels defaults
conda config --add channels conda-forge
conda config --add channels bioconda

MetaMeta:

conda install metameta=1.2.0

• All other tools and dependencies are installed in their own environment automatically on the first run (with --use-conda parameter active).

Alternatively, install MetaMeta in a separated environment (named "metametaenv") with the command:

conda create -n metametaenv metameta=1.2.0	
source activate metametaenv # Command to activate the environment. To deactivate use "source deactivate"

Run:

Create a configuration file (yourconfig.yaml) with the required fields (workdir, dbdir and samples):

workdir: "/home/user/folder/results/"
dbdir: "/home/user/folder/databases/"
samples:
  sample_name_1:
     fq1: "/home/user/folder/reads/file.1.fq"
     fq2: "/home/user/folder/reads/file.2.fq"

• All paths set on this file are relative to the workdir (if not absolute)

Check rules and output files:

metameta --configfile yourconfig.yaml -np

Run MetaMeta:

metameta --configfile yourconfig.yaml --use-conda --keep-going --cores 24

• Alternatively, make a copy of the configuration file for the complete set of parameters cp ~/miniconda3/opt/metameta/config/example_complete.yaml yourconfig.yaml
• The number of --cores is the total amount avaiable for the pipeline. Number of specific threads for the tools should be set on the configuration file (yourconfig.yaml) with the parameter threads
• On the first run MetaMeta will download and install the configured tools as well as the database files (archaea_bacteria_201503 by default - see below) necessary for each tool.

Pre-configured databases:

Available databases:

Info	Date	metameta database name
Archaea + Bacteria - RefSeq Complete Genomes	2015-03	archaea_bacteria_201503
Fungal + Viral - RefSeq Complete Genomes	2017-09	fungi_viral_201709

Database availability per tool:

database	clark	dudes	gottcha	kaiju	kraken	motus
archaea_bacteria_201503	Yes	Yes	Yes	Yes	Yes	Yes
fungi_viral_201709	Yes	Yes	No	Yes	Yes	No

Running sample data:

cd ~/miniconda3/opt/metameta/

Pre-configured Archaea and Bacteria database:

./metameta --configfile sampledata/sample_data_archaea_bacteria.yaml --use-conda --keep-going --cores 6

Custom database (some viral reference genomes):

./metameta --configfile sampledata/sample_data_custom_viral.yaml --use-conda --keep-going --cores 6

Results:

cd sampledata/results/ 

Running MetaMeta on a cluster environment:

Make a copy of cluster configuration file:

cp ~/miniconda3/opt/metameta/config/cluster.json yourcluster.json

Edit the file with your cluster specifications (threads, partitions, cpu/memory, etc) for each rule.

Run MetaMeta (slurm example):

metameta --configfile yourconfig.yaml --keep-going --use-conda -j 999 --cluster-config yourcluster.json --cluster "sbatch --job-name {cluster.job-name} --output {cluster.output} --partition {cluster.partition} --nodes {cluster.nodes} --cpus-per-task {cluster.cpus-per-task} --mem {cluster.mem} --time {cluster.time}"

• you can change the cluster command (sbatch) and adapt them to your cluster system.

Custom databases:

MetaMeta uses by default Archaea and Bacteria sequences as reference database (archaea_bacteria_201503 - see below). Additionaly MetaMeta allows the creation of custom database.

First select which databses should be used on the configuration file:

databases:
  - archaea_bacteria_201503
  - custom_db

• all samples will run agains the "archaea_bacteria_201503" and the new "custom_db" databases

Second, create an entry with the path to the sequences that should be added to the custom database:

custom_db:
    clark: "sampledata/database/"
    dudes: "sampledata/database/"
    kaiju: "sampledata/database/"
    kraken: "sampledata/database/"

• clark and dudes require one or more fasta files (extension .fna) with the accession.version identifier after the header ">" (e.g. ">NC_001998.1 Guinea pig Chlamydia phage, complete genome")
• kaiju requires one or more GenBank flat file (extension .gbff)
• kraken requires one or more fasta files (extension .fna) with the gi identifier on the header (e.g. ">gi|9632287|ref|NC_001998.1| Guinea pig Chlamydia phage, complete genome")

MetaMeta will compile the "custom_db" on the first run and use it as a database. After finished it is possible to delete de database definition from the configuration file for the following runs.

Creating a custom database based on NCBI genomes:

It is possible to create a custom database based on the set of genomes from NCBI

Download the genome_updater script:

git clone https://github.com/pirovc/genome_updater

Download the desired database: Example -> All fungi genomes available on refseq, fasta and GenBank formats with 6 threads:

./genome_updater.sh -d "refseq" -g "fungi" -f "genomic.fna.gz,genomic.gbff.gz" -t 6 -o fungi_genomes/
mkdir -p custom_fungi_db/clark_dudes/ custom_fungi_db/kaiju/ custom_fungi_db/kraken/

Extract files: clark and dudes:

zcat fungi_genomes/files/*.fna.gz > custom_fungi_db/clark_dudes/fungi_genomes.fna

kaiju:

zcat fungi_genomes/files/*.gbff.gz > custom_fungi_db/kaiju/fungi_genomes.gbff

kraken (with header conversion to GI, old NCBI style):

zcat fungi_genomes/files/*.fna.gz | awk '{if(substr($0, 0, 1)==">"){sep=index($0," ");acc=substr($0,2,sep-2);header=substr($0,sep+1); cmd="wget -qO - \"https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nucleotide&id="acc"&rettype=gi\""; cmd | getline gi; close(cmd); print ">gi|" gi "|ref|" acc "| " header }else{ print $0 }}' > custom_fungi_db/kraken/fungi_genomes.fna

Add entry on the configuration file:

databases:
  - new_custom_fungi_db

Finally, add the path for each set of reference sequences on the configuration file:

new_custom_fungi_db:
    clark: "custom_fungi_db/clark_dudes/"
    dudes: "custom_fungi_db/clark_dudes/"
    kaiju: "custom_fungi_db/kaiju/"
    kraken: "custom_fungi_db/kraken/"	

On the first run MetaMeta will compile the "new_custom_fungi_db" database for each configured tool. After finished it is possible to delete de database definition from the configuration file for the following runs.

Pre-install a complete environment:

wget https://raw.githubusercontent.com/pirovc/metameta/master/envs/metameta_complete.yaml
conda env create -f metameta_complete.yaml
source activate metametaenv_complete

Merging final results:

To merge final results from many samples into one final tabular file:

~/miniconda3/opt/metameta/scripts/merge_final_profiles.sh workdir/samples_*/metametamerge/database/final.metametamerge.profile.out

Folder structure:

MetaMeta can run several tools with several samples against several databases. The files on the working directory and database directory are organized in the structure below:

WORKDIR:
	SAMPLE_1/
		TOOL_1/ (*)
			DB_1/
			DB_2/
			...
		TOOL_2/ (*)
			...
		PROFILES/
			DB_1/
				TOOL_1.profile.out
				TOOL_2.profile.out
				...
			DB_2/
				...
		METAMETAMERGE/
			DB_1/
				FINAL_PROFILE.out
				FINAL_PROFILE_KRONA.html
			DB_2/
				...
		LOG/
			DB_1/
			DB_2/
			...
		READS/ (*)
			TOOL_1.1.fq
			TOOL_1.2.fq
			TOOL_2.1.fq
			TOOL_2.2.fq
			...
	SAMPLE_2/
		...
	CLUSTERLOG/ (**)

DBDIR:
	DB_1/
		TOOL_1_DB/
		TOOL_2_DB/
		...
		TOOL_1.dbprofile.out
		TOOL_2.dbprofile.out
		...
		LOG/
	DB_2/
		...
	TAXONOMY/
		LOG/

(*) removed when keepfiles=0 (**) only when running on cluster mode

Adding a new tool:

MetaMeta integrates profiling and binning tools and it has 6 pre-configured tools (clark, dudes, gottcha, kaiju, kraken and motus). New tools are required to use the NCBI Taxonomy structure and nomenclature/identifiers to be added to the pipeline. MetaMeta accepts BioBoxes format directly (https://github.com/bioboxes/rfc/tree/master/data-format) or a .tsv file in the following format:

• Profiling: rank, taxon name or taxid, abundance

Example:

genus   Methanospirillum        0.0029
genus   Thermus 0.0029
genus   568394      0.0029
species Arthrobacter sp. FB24   0.0835
species 195      0.0582
species Mycoplasma gallisepticum        0.0536

• Binning: readid, taxon name or taxid, lenght of sequence assigned

Example:

M2|S1|R140      354     201
M2|S1|R142      195     201
M2|S1|R145      457425  201
M2|S1|R146      562     201
M2|S1|R147      1245471 201
M2|S1|R150      354     201

MetaMeta pipeline uses Snakemake. To add a new tool to the pipeline it is necessary to create two main files described below. Replace 'newtool' with the tool identifier (lower case, no spaces, no special chars):

tools/newtool.sm -> specifies how to execute the tool
	Rules:
	- newtool_run_1[..n] -> one or more rules necessary to run the tool
	- newtool_rpt -> final rule that should output a file newtool.profile.out in an accepted output format (described above)
	
tools/newtool_db_custom.sm -> specifies how to download/compile the database/references
	Rules:
	- newtool_db_custom_1[..n] -> one or more rules necessary to compile the database.
	- newtool_db_custom_profile -> this rule generates automatically the database profile. It should have as an output a file (newtool.dbaccession.out) with the accession version identifier for all sequences used in the database.
	- newtool_db_custom_check -> rule to check the required database files. It should have as an input all mandatory files that should be present to the database work properly.

• Template files can be found inside the folder tools/template. Once the two files are inside the tools folder, it is necessary to add the tool identifier to the YAML configuration file.

Changelog:

v1.2.0)

• Updated to Snakemake 4.3.0 (from 3.9.1)
• Bug fixes on custom database creation and database profile generation
• Centralized taxonomy download (once for all tools, kept on dbdir:taxonomy/)
• Updated tools: kaiju 1.0 -> 1.4.5, dudes 0.07 -> 0.08, spades 3.9.0 -> 3.11.1
• Addition of new pre-configured databases: fungal_viral_201709
• Multiple pre-configured databases support
• Several fixes on custom database creation

v1.1.1) Bug fixes parsing output files for kraken and kaiju

v1.1) Support single and paired-end reads, multiple and custom databases, krona integration