Metagenomics virus detection (MetaV): rapid detection and classification of viruses in metagenomics sequencing

1. Introduction

MetaV is a command-line-interface program, which is used to rapidly identify and classify viral sequences from metagenomic sequencing data. metav is developed via Python 3, and can be run on Linux systems and deployed to the cloud.

2. Download and install

pip install metav
metav -h

3. Software dependencies

The running of metav relies on these softwares:

• Trimmomatic (version >=0.39), which is used to remove the contamination from adapter primer.

• Bowtie2 (version >=2.3.5), which is used to remove the contamination from host genome.

• Trinity (version >=2.15.1), in the second sub-pipeline of metav, the Trinity is used to splice reads to contigs. Note, the running of Trinity relies on Bowtie2, jellyfish, samtools and salmon, and they can be easily installed,

sudo apt-get install bowtie2
sudo apt-get install jellyfish
sudo apt install samtools
sudo apt install salmon

• diamond (version >=2.0.9), the diamond is used to map reads (or contigs) to proteins.

Note, The four dependencies (Trimmomatic, Bowtie2, Trinity and diamond) need to be installed manually by users in advance and be added to PATH (system or user).

4. Database dependencies

4.1. prepare host database

The host database is used to remove contamination from host genome. How to prepare a host database?

(1) download the genomic data of host with *.fasta format.

(2) creat the host database using Bowtie2 software, for example, bowtie2-build /home/zzj/host_db/host_genome.fna /home/zzj/host_db/host_genome. It then generates six files, which starts with "host_genome" and suffix are '.1.bt2', '.2.bt2', '.3.bt2', '.4.bt2', '.rev.1.bt2', and '.rev.2.bt2'.

Next, you need to fill in the path /home/zzj/host_db/host_genome into file profiles.xml. Note, the path /home/zzj/host_db/host_genome is not a directory!

(3) metav also supports multiplehost databases, please use , to separate these path in file profiles.xml, for example, /home/zzj/host_db/host_genome1, /home/zzj/host_db/host_genome2.

Tip, different samples may come from different hosts, please adjust them in file profiles.xml in time.

4.2. prepare viral nr database

The viral nr database was used to identity viral components from sequenced reads. How to prepare a viral nr database?

(1) firstly, download the refseqs of viral proteins (amino acid, *.1.protein.faa.gz) from viral database of ncbi refseqs, besides, please also download *.protein.gpff.gz containning the taxonomic information of these sequences. Note, the format of file *.1.protein.faa.gz is fasta.

(2) next, unzip *.1.protein.faa.gz and rename to ViralProtein.fasta, then creat the viral nr database using diamond software, for example, diamond makedb -p 10 --in /home/zzj/nr/ViralProtein.fasta --db /home/zzj/nr/ViralProtein.dmnd. Then, fill in the path /home/zzj/nr/ViralProtein.dmnd into file profiles.xml.

(3) then, extract the viral taxonomy information from file *.protein.gpff.gz , which is used to classfy viral reads. This repository provides the taxonomy_information_2021-05-20.txt made by ourselves, in which the accession is consistent with the file ViralProtein.fasta. If you want to add some information, please keep it in the same format (four columns, don't change the name of column). Finally, fill in the path of taxonomy information file into the file profiles.xml.

Tip, the viral nr database generally does not need to be replaced in the short term.

5. Configuration of dependencies

In order to manage the parameters of dependent softwares and databases convenienty, the profiles.xml file is used to record their configuration.

the template of profiles.xml is provided in the github repository, please note,

• currently version of metav only supports the sequenced data from Illumina platform.

• the paths of these databases in file profiles.xml need to be adjusted with reference to your computer, databases paths in profiles.xml we provided were just some examples. Note, they have to be absolute path, not relative path.

• the parameters of software in profiles.xml generally does not need to be modified because they are suitable in most cases. Note, the path of adapters file needs to be modified, see field ILLUMINACLIP:/home/zzj/software/bioinfo/trimmomatic/adapters/merge_adapter.fas in setting of trimmomatic in profiles.xml. The path /home/zzj/software/bioinfo/trimmomatic/adapters/merge_adapter.fas here is only an example, adapter file is generally in the adapters folder of the installation directory of trimmomatic software, or you can make this file yourself, just fill in the corresponding absolute path here.

Tip, in general, these parameters only need to be configured once in the first running, except for the host database used to filter contamination of host genome.

6. Getting help

Users can view the help documentation by entering metav -h or metav --help .

7. Example of usage

• if reads are from paired-end sequencing:

metav -pe -i1 reads_R1.fq -i2 reads_R2.fq -xml profiles.xml -r1 -r2 -t 8 -o outdir

• if reads are from single-end sequencing:

metav -se -u reads.fq -xml profiles.xml -r1 -r2 -t 8 -o outdir

Tip

• metav is also supported to run one of -r1 and -r2.

• if -r2 is used, the output directory behind -o have to be absolute path.

• if an error is displayed, please check the input parameters and XML file.

8. Output results

8.1. input-parameter.txt

the output file input-parameter.txt recorded the input parameters in command-line interface.

the used parameters of metav in command-line interface.

pair_end:	True
single_end:	False
sub-pipeline 1:	True
sub-pipeline 2:	True
forward_reads:	/home/zzj/datas/test/reads_1.fq
reverse_reads:	/home/zzj/datas/test/reads_2.fq
unpaired:	None
qualities:	phred33
set_file:	/home/zzj/datas/test/profiles.xml
length_threshold:	10.0
identity_threshold:	20.0
e-value:	['1e-6', '1e-3', '1e-1']
thread:	8
outdir:	/home/zzj/datas/test/out6

8.2. directory pipeline1

the directory pipeline1 contains intermediate results and finally_result from sub-pipeline 1 (reads → nr database).

In the example, three thresholds (1e-6, 1e-3 and 1e-1) of e-value are used to filter the output diamond program. Thus, three corresponding sub-directories is created and used to store results.

The meanings of directory name with e-value in pipeline1 are as follows,

The hierarchy is same in all three sub-directories with e-value. For example, in the directory hit_summary of the directory lower_1e-6, metav provides a summary file (hit_reads_taxonomy_information.txt) with taxonomy information.

What's more, metav counts these hit reads according to order, family and strain(organism) and provides three *.csv summary files.

In particular, metav extract all hit reads sequences (*fasta format) according to the hierarchical relationship of order, family and strain(organism). These hit reads sequences are stored in directory hit_reads_seq.

8.3. directory pipeline2

the directory pipeline2 contains intermediate results and finally_result from sub-pipeline 2 (reads → contigs → nr database). The hierarchy of the directory of output results is the same as directory pipeline1.

However, the output in directory finally_result of the directory pipeline2 are hit contigs, not reads.

The meanings of directory name with e-value in pipeline2 are as follows,

In the directory hit_summary of each sub-directory with e-value, the sequences and summary information of hit contigs are provided, and these hit contigs sequences are stored in directory hit_contigs_seq.

9. Functional expansion

metav was originally designed to detect and count the viral composition in metagenomics-sequencing-data, but it's flexible and not limited to viruses.

In fact, the viral nr database can be replaced by protein databases of other pathogenic, for example, bacteria, pathogenic fungi. These nr database cam be download from database of ncbi refseqs. In a word, metav can detect and count other pathogens of metagenomics-sequencing-data by using the corresponding nr database and taxonomy information file.

10. Bug report

metav was test on Ubuntu 16.04 and Ubuntu 20.02, which can work well. If you run into a problem or find a bug, please contact us.

Github issues or send email to zjzhou@hnu.edu.cn.