Manual for Sequence Search Tool for Antimicrobial Resistance (SSTAR), version 1.1.01

Table of Contents

• Table of Contents
  • Introduction
  • Two SSTAR versions
  • Obtaining and installing SSTAR dependencies
  • AR gene databases
  • Input data
  • Running SSTAR
  • BLAST output file produced by SSTAR
  • Planned features
  • FAQ
  • Citing SSTAR
  • Contact

Introduction

SSTAR enables fast and accurate antimicrobial resistance (AR) surveillance from Whole Genome Sequencing (WGS) data. It is able to identify known AR genes and detect putative new variants as well as truncated genes due to internal stop codons. SSTAR can also report modifications and/or truncations in outer membrane porin genes.

Two SSTAR versions

In your downloaded archive you will find two different SSTAR versions

1. SSTAR.jar is for Linux and OS X
2. SSTAR_windows.jar is for Windows

JAR and Java files

• Each SSTAR version is available as an executable JAR file. Double click on this file and SSTAR will pop up on your screen
• Raw Java files are also available for those who want to explore the source code
• SSTAR was successfully tested under Windows 7 and Windows 10, OS X 10.9.5, and Ubuntu 14.04 LTS

Obtaining and installing SSTAR dependencies

SSTAR combines a standalone BLASTN with a Java interface, which operates under Windows and Unix systems. In order to run SSTAR under Windows you need Java Runtime Environment (JRE) 6 or newer and standalone BLAST 2.2.29+, BLAST 2.2.30+ or BLAST 2.2.31+. SSTAR for Linux and OS X is compatible with JRE6 or newer and all BLAST+ versions.

For Windows users

• BLAST+ needs to be installed in C:\Program Files\NCBI which is the default location when following the BLAST installation steps in the installation wizard. SSTAR for Windows is currently compatible with BLAST versions: 2.2.29+, 2.2.30+, 2.2.31+, 2.5.0+ and 2.6.0+

For OS X and Linux users

• BLAST+ can be installed anywhere and SSTAR is compatible with all BLAST+ versions. BLAST+ needs to be added to your PATH variable. Additionally, please add (using the "echo" command) two important lines to your .bash_profile or .profile depending on your shell, and operating system

    echo 'export BLASTN="blastn"' >> $HOME/.bash_profile
    echo 'export MAKEBLASTDB="makeblastdb"' >> $HOME/.bash_profile

Load your .bash_profile or .profile into the current shell or command prompt

    source .bash_profile
    OR
    source .profile

You can check if your variables were exported correctly using the echo command and the commands should return "blastn" and "makeblastdb"

    echo $BLASTN
    echo $MAKEBLASTDB

For all users

• Java can be downloaded from the Oracle website at: https://java.com/en/download/index.jsp
• A BLAST+ copy and easy installation guide of standalone BLAST can be retrieved from: http://www.blaststation.com/freestuff/en/howtoNCBIBlastWin.html

AR gene databases

We have generated an AR gene database from Resfinder and ARG-ANNOT data, called resGANNOT. Nucleotide genes were clustered using CD-HIT at 90% sequence similarity and redundant entries were discarded. resGANNOT is maintained by Nicholas Vlachos (NVX4@cdc.gov)

Input data

One needs two input files for SSTAR to run: A microbial genome assembly and AR gene file, both in FASTA format. SSTAR is developed in a certain way so it can handle the ‘SRST2 database header’ format. Two AR database files are included with SSTAR, a SRST2 modified ARG-ANNOT database and a SRST2 modified Resfinder database. This format is specified below.

The AR gene file header format

The format of each AR gene FASTA header is structured like the below example:

92__CMY_Bla__CMY-37__402

The AR gene family is between the first and second double underscore (CMY_Bla) and the variant is between the second and third double underscore (CMY-37). The first number (92) is a unique identifier for each AR gene group. The last number (402) is a unique identifier for each single variant. The other information in the header (right of the space) is ignored by SSTAR and not shown in this manual example.

Users who want to use different AR databases (ResFinder or custom databases) need to make sure the headers have the exact same structure as the SRST2 header format.

Running SSTAR

SSTAR contains an easy interface with currently only four buttons. The top two buttons are for uploading the genome assembly file and the AR gene database file. Both files need to be in FASTA format. One needs to enter a sequence similarity percentage value that serves as cut-off for detecting potential new variants of AR genes. A value between 80 and 99% is recommended. The ‘Identify resistance genes’ button starts the actual AR gene annotation process. The genes will be listed in the the second output window.

AR gene output

The second output window displays the AR resistance genes and porins that are identified on your input genome assembly. Each output line contains five fields, separated by a tab:

1. The AR gene name
2. The contig, scaffold or chromosome where the AR gene is located
3. Sequence similarity
4. Alignment length
5. AR gene length

First, SSTAR lists the potential new alleles or variants at the top of the window. These are the variants that share between X% and 99.99% sequence similarity with an AR gene in the used database. The user is free to pick a value for X. Below the potential new variants SSTAR lists the AR genes that share 100% sequence similarity with AR genes in the used database. The alignment length shows the user how much of the AR database gene is found on your genome. In other words, when the alignment length equals the gene length one identified the full AR gene with 100% sequence similarity.

The export button will export the data as a tab delimited file and can then be opened in a spreadsheet, such as Microsoft Excel. The file is stored in the same directory as the input genome assembly file

Detecting new and truncated AR gene variants

The bottom output window will show putative new variants and truncated enzymes in protein space. The protein sequences can be exported to a plain text file, in FASTA format, using the export button. The file is saved in the same directory as the input genome assembly file. The protein file can be used with BLASTP against the NR database of NCBI for detecting new variants. Potential novel beta-lactamase proteins can be submitted to the NCBI (http://www.ncbi.nlm.nih.gov/pathogens/submit_beta_lactamase/) for verification. Translated start codons (Methionines, M) are capitalized so the user gets a better idea where the protein starts. Not all proteins start with an M, however in that case SSTAR will report the ORF with the fewest internal stop codons. When a protein sequence contains an internal stop codon it will be flagged underneath the FASTA header of that particular protein. This makes the FASTA file invalid and forces the user to remove that sequence from the file. Protein sequences with internal stop codons are otherwise easily missed and misinterpreted as putative new variants of an AR gene group.

Detecting modified and truncated outer membrane porin sequences

The bottom output window will also show modified and/or truncated outer membrane porins (OMPs). We have included OmpK35, OmpK36 and OmpK37 from Klebsiella pneumoniae, OmpC and OmpF from Escherichia coli, OmpC and OmpF from Enterobacter cloacae and Omp35 and Omp36 from Enterobacter earogenes. When a porin protein sequence contains an internal stop codon it will be flagged underneath the FASTA header of that particular porin as truncated.

BLAST output file produced by SSTAR

The results that are generated by SSTAR are shown in the graphical interface as explained in the previous section, however SSTAR is also producing a raw BLASTN file in the same folder as your input data:

The BLASTN file is in tabular form and each line represents an allele that was part of your query database. A line contains 13 fields and are briefly described here:

1. Query (An antimicrobial gene allele)
2. Target sequence (A contig, scaffold or chromosome of your input genome file)
3. % ID
4. Alignment length
5. Number of mismatches
6. Number of gap opens
7. Start position in query
8. End position in query
9. Start position in target
10. End position in target
11. E-value
12. Bit score
13. Query length

Planned features

1. Multiple genome assembly file upload <-- please see c-SSTAR
2. Compatibility with other BLAST+ versions (Windows version only, Unix version works with all BLAST versions already)
3. An eraser button for all three output windows
4. A method to use two AR gene databases (ResFinder and ARG-ANNOT) simultaneously <-- please use our combined AR gene database "resGANNOT"
5. Add HMMER functionality for detecting new genes
6. Build command line version of SSTAR for Linux and OS X <-- please see c-SSTAR

FAQ

1. SSTAR doesn’t like certain CLC Genomics Workbench assembly files. Why?
   

   CLC generates FASTA headers that contain spaces.
   Since all information immediately right of the first space is disregarded it can cause each contig/scaffold to have the exact same name as all the other contigs/scaffolds. This will cause issues during downstream analyses since SSTAR won’t be able to discern between the different contigs.
   The issue can be fixed by converting the FASTA headers to a “SSTAR-friendly” format using the split_fasta_header_on_space.pl script in the Scripts folder.

Citing SSTAR

Please cite our paper in mSphere:
de Man TJB, Limbago BM. 2016. SSTAR, a stand-alone easy-to-use antimicrobial resistance gene predictor. mSphere 1(1): e00050-15

When using the ARG-ANNOT database please also cite:
Gupta SK, Padmanabhan BR, Diene SM, Lopez-Rojas R, Kempf M, Landraud L, Rolain J-M. 2014. ARG-ANNOT (Antibiotic Resistance Gene-ANNOTation), a new bioinformatic tool to discover antibiotic resistance genes in bacterial genomes. Antimicrobial Agents and Chemotherapy 58:212–220.

When using the ResFinder database please also cite:
Zankari E, Hasman H, Cosentino S, Vestergaard M, Rasmussen S, Lund O, Aarestrup F, Larsen MV. 2012. Identification of acquired antimicrobial resistance genes. Journal of Antimicrobial Chemotherapy 67:2640–2644.

Contact

For assistance, feedback or suggestions please contact Tom de Man via tjb.deman@gmail.com