Conterminator

Detection of contamination in nucleotide and protein sequence sets

Conterminator is an efficient method for detecting incorrectly labeled sequences across kingdoms by an exhaustive all-against-all sequence comparison. It is a free open-source GPLv3-licensed software for Linux and macOS, and is developed on top of modules provided by MMseqs2.

Terminating contamination: large-scale search identifies more than 2,000,000 contaminated entries in GenBank. Genome Biology, doi: 10.1186/s13059-020-02023-1 (2020)

Install

Conterminator requires a 64-bit Linux system (check that uname -a | grep x86_64 | wc -l is greater than 0) with at least a SSE4.1 instruction set (check that cat /proc/cpuinfo | grep sse4_1 | wc -l yields an output greater than 0).

# SSE4.1
wget https://mmseqs.com/conterminator/conterminator-linux-sse41.tar.gz; tar xvfz conterminator-linux-sse41.tar.gz; export PATH=$(pwd)/conterminator/:$PATH
# AVX2
wget https://mmseqs.com/conterminator/conterminator-linux-avx2.tar.gz; tar xvfz conterminator-linux-avx2.tar.gz; export PATH=$(pwd)/conterminator/:$PATH
# conda
conda install -c bioconda conterminator

Getting started

Conterminator computes ungapped local alignments of all provided sequences and reports contamination across user-specifed specified taxa; by default this is done at the kingdom level.

Conterminator requires two input files: (1) a FASTA file containing all sequences (example/dna.fna/example/prots.faa) and (2) a mappingFile (example/dna.mapping/examples/prots.mapping), which maps FASTA identfiers to NCBI taxon identfiers. The program produces two output files with prefix (${RESULT_PREFIX}). More details are described in the Results section below.

To process nucleotide sequences use the following command:

conterminator dna example/dna.fna example/dna.mapping ${RESULT_PREFIX} tmp     

Protein sequences can be processed as following:

conterminator protein example/prots.faa example/prots.mapping ${RESULT_PREFIX} tmp  

Mapping file

Conterminator needs a mapping file, which assigns each fasta identifier to a taxonomical identifier. The mapping file consists of two tab-delimited columns, (1) fasta identifier and (2) [NCBI taxonomy identifier] (taxonomy ID) (https://www.ncbi.nlm.nih.gov/taxonomy). By default, Conterminator takes the text up to the first blank space as the fasta identifier. However, with GenBank, Tremble, Swissprot, Conterminator extracts out only the unique identifier mapped to the taxonomy ID.

Example for detecting contamination in the NT database:

blastdbcmd -db nt -entry all > nt.fna
blastdbcmd -db nt -entry all -outfmt "%a %T" > nt.fna.taxidmapping
conterminator dna nt.fna nt.fna.taxidmapping nt.result tmp

Result

Conterminator produces two result files (1) ${RESULT_PREFIX}_conterm_prediction and (2) ${RESULT_PREFIX}_all. The {RESULT_PREFIX}_conterm_prediction text file contains the predicted contamination. The file is TSV-seperated containing the following columns:

1.) Numeric identifier
2.) Contaminated identifier
3.) Kingdom (default: 0: Bacteria&Archaea, 1: Fungi, 2: Metazoa, 3: Viridiplantae, 4: Other Eukaryotes)
4.) Species name
5.) Alignment start
6.) Alignment end
7.) Corrected contig length (length between flanking Ns)
8.) Identifier of the longest contaminating sequence
9.) Kingdom of the longest contaminating sequence
10.) Species name of the longest contaminating sequence
11.) Length of the longest contaminating sequence
12.) Count how often sequences from the contaminating kingdom align

Be aware that the result file may contain any contaminated identifier multiple times if multiple alignments were detected.

The {RESULT_PREFIX}_all contains the information for all alignments that were used to predict contamination.

1.) Numeric identifier
2.) Sequence identifier
3.) Alignment start
4.) Alignment end
5.) Corrected contig length (length between flanking Ns)
6.) Total sequence length
7.) Kingdom (default: 0: Bacteria&Archaea, 1: Fungi, 2: Metazoa, 3: Viridiplantae, 4: Other Eukaryotes)
8.) Species name 

Important Parameters

--kingdom

This parameter specifies which taxons across which contaminations should be considered. Each taxon definition is seperated by a , e.g. to search for contamination between bacteria and human use --kingdom 2,9606. It is also possible to use more advanced expressions for contamination rules, through the following operators:

! NEGATION 
|| OR  
&& AND 

The default rule is as follows:

2||2157,4751,33208,33090,2759&&!4751&&!33208&&!33090   

This searches for contamination between the following taxa:

2||2157  # Bacteria OR Archaea 
4751     # Fungi
33208    # Metazoa
33090    # Viridiplantae  
2759&&!4751&&!33208&&!33090 # Eukaryota without Fungi Metazoa and Viridiplantae

[OPTIONAL] Install by Compilation

Users can install Conterminator using the commands specified above. However, Conterminator can also be installed by compiling directly from the source code using the following commands.

git clone --recursive https://github.com/martin-steinegger/conterminator 
mkdir conterminator/build && cd conterminator/build
cmake -DCMAKE_BUILD_TYPE=RELEASE -DCMAKE_INSTALL_PREFIX=. ..
make -j 4
make install
export PATH=$(pwd)/bin/:$PATH