Published online 29 October 2019 Nucleic Acids Research, 2020, Vol. 48, Database issue D511–D516 doi: 10.1093/nar/gkz931 BacFITBase: a database to assess the relevance of bacterial genes during host infection 1,† 2,† 2,3,4,5,* Javier Macho Rendón , Benjamin Lang , Gian Gaetano Tartaglia and Marc Torrent Burgas 1,* 1 Systems Biology of Infection Lab, Department of Biochemistry and Molecular Biology, Biosciences Faculty, Universitat Autònoma de Barcelona, 08193 Cerdanyola del Vallès, Spain, 2 Centre for Genomic Regulation (CRG), The Barcelona Institute for Science and Technology, Dr. Aiguader 88, 08003 Barcelona, Spain, 3 ICREA, 23 Passeig Lluis Companys 08010 and Universitat Pompeu Fabra (UPF), 08003 Barcelona, Spain, 4 Department of Biology ‘Charles Darwin’, Sapienza University of Rome, P.le A. Moro 5, Rome 00185, Italy and 5 Department of Neuroscience and Brain Technologies, Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genoa, Italy Received August 15, 2019; Revised October 04, 2019; Editorial Decision October 06, 2019; Accepted October 07, 2019 ABSTRACT vitro studies (3). This is a major drawback for antimicro- bial target discovery as many in vitro false negatives are dis- Bacterial infections have been on the rise world- regarded for further testing. Therefore, it is crucial to un- wide in recent years and have a considerable im- derstand how bacterial infection develops in vivo and which pact on human well-being in terms of attributable bacterial genes are required to infect a host. deaths and disability-adjusted life years. Yet many BacFITBase is a manually curated database of bacterial mechanisms underlying bacterial pathogenesis are genes that collates in vivo information on their relevance still poorly understood. Here, we introduce the Bac- during host infection, as measured by transposon mutage- FITBase database for the systematic characterization nesis. Transposon mutagenesis experiments allow the mea- of bacterial proteins relevant for host infection aimed surement of fitness values for individual genes, allowing to enable the identification of new antibiotic targets. us to assess which genes are fundamental to infect a spe- BacFITBase is manually curated and contains more cific host organism (4). To address the contribution of a bacterial gene to infection, its fitness is measured through than 90 000 entries with information on the contribu- genome-wide transposon mutagenesis coupled with next- tion of individual genes to bacterial fitness under in generation sequencing (Tn-seq) (5). Briefly, mutations tar- vivo infection conditions in a range of host species. geting virtually all genes in the bacterial genome are gener- The data were collected from 15 different studies in ated by random insertion of transposons. Afterward, these which transposon mutagenesis was performed, in- mutants with randomly inactivated genes are grown in cul- cluding top-priority pathogens such as Acinetobac- ture medium (input pool), inoculated in a host organism ter baumannii and Campylobacter jejuni, for both and finally recovered after infection (output pool). Ge- of which increasing antibiotic resistance has been nomic DNA from the input and output pool is extracted, reported. Overall, BacFITBase includes information and transposon insertion site junctions are amplified and on 15 pathogenic bacteria and 5 host vertebrates quantified by next-generation sequencing. across 10 different tissues. It is freely available at BacFITBase provides a common framework and easy access to fitness data for individual bacterial genes dur- www.tartaglialab.com/bacfitbase. ing infection. It now covers the infection processes of 15 pathogenic bacteria in 5 model vertebrates across 10 dif- INTRODUCTION ferent tissues (Table 1). A tutorial section provides a de- tailed step-by-step description of how to search and browse The development of new antimicrobial therapies relies heav- our data. This resource is available at www.tartaglialab.com/ ily on our understanding of the mechanisms of bacterial bacfitbase to help the research community in the systematic infection. Bacterial proteins are responsible for rewiring a characterization of bacterial proteins involved in host infec- myriad of biochemical processes essential for the efficient tion. propagation of the pathogen (1,2). Recently, we showed that bacterial fitness in vivo does not correlate with data from in * To whom correspondence should be addressed. Tel: +34 93 581 4147; Email: marc.torrent@uab.cat Correspondence may also be addressed to Gian G. Tartaglia. Tel: +34 93 316 0116; Email: gian.tartaglia@crg.eu † The authors wish it to be known that, in their opinion, the first two authors should be regarded as Joint First Authors. C The Author(s) 2019. Published by Oxford University Press on behalf of Nucleic Acids Research. This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact journals.permissions@oup.com D512 Nucleic Acids Research, 2020, Vol. 48, Database issue Table 1. List of all studies included in the BacFITBase Pathogen Host organism Tissue Reference Salmonella enterica Serovar Typhimurium ST4/74 Bos taurus (cow) Ileal mucosa (6) Sus scrofa (pig) Colonic mucosa Gallus gallus (chicken) Cecum Haemophilus influenzae Rd KW20 Mus musculus (mouse) Lung (7) Streptococcus pyogenes M1 5448 M. musculus (mouse) Skin (8) Porphyromonas gingivalis ATCC 33277 M. musculus (mouse) Skin (9) Escherichia coli CFT073 M. musculus (mouse) Spleen (10) Mycobacterium avium subsp. Paratuberculosis K10 M. musculus (mouse) Spleen (11) Escherichia coli M12 M. musculus (mouse) Spleen and mammary gland (12) E. coli O157:H7 B. taurus (cow) Feces (13) Vibrio cholerae O1 biovar El Tor str. N16961 Oryctolagus cuniculus (rabbit) Small intestine (14) Campylobacter jejuni subsp. jejuni 81–176 M. musculus (mouse) Cecum (15) Klebsiella pneumoniae subsp. pneumoniae M. musculus (mouse) Lung (16) pneumoniae ATCC 43816 KPPR1 Acinetobacter baumannii ATCC 17978 M. musculus (mouse) Lung (17) Salmonella enterica Serovar Typhimurium SL1344 M. musculus (mouse) Liver and spleen (18) Serratia marcescens Strain UMH9 M. musculus (mouse) Spleen (19) Vibrio parahaemolyticus RIMD 2210633 O. cuniculus (rabbit) Small intestine (20) MATERIALS AND METHODS Fitness scores and z-scores We collected data from publicly available transposon mu- tagenesis experiments containing either raw input/output read counts or fitness scores for all mutant genes available (Table 1) (6–20). The fitness score of a gene is calculated as the ratio of the normalized frequencies of input/output read counts. Reads for each transposon insertion site are normalized to the total number of reads obtained from a sample according to the following equation: Cia /Ni Fa = C0a /N0 where F a is the fitness score for gene a, C0a and Cia are the number of reads for a gene a before and after infection and N0 and Ni are the number of total reads before and after in- Figure 1. Histogram representing the z-score distribution among all stud- ies included in the BacFITBase database. A fitness z-score < 0 indicates fection, respectively. In order to normalize the fitness scores that a given mutation is more detrimental than the average mutation during to allow comparison between different studies, the corre- infection, while a raw fitness score < 1 indicates that a given mutant is un- sponding z-scores were calculated for each individual ex- derrepresented in the output pool. As we can observe in the histogram, the periment (Figure 1). To assess the significance of a bacte- extreme values in the fitness z-score distribution are shifted toward nega- tive values. Under-represented instances give strong evidence that this par- rial gene’s fitness impact, P-values were calculated using a ticular gene is relevant in the infection process, as transposon insertion is two-tailed one-sample Student’s t-test on the distribution of typically associated with a loss of function phenotype. fitness scores within each study provided that raw data were available. Otherwise, the P-values reported in the original study are shown. BLAST search Technical aspects The NCBI BLAST suite version 2.9.0+ (March 2019) (21) BacFITBase was built using PHP on an Apache web server is used to search by sequence similarity. The BLASTP pro- with a MySQL database backend. BacFITBase stores no gram is used for amino acid sequences, and BLASTX for user data, except for the anonymous caching of BLAST nucleic acid (coding) sequences. BLAST search results are search results for a given sequence in order to greatly speed cached for each unique sequence, which means that re- up repeated searches. The open-source Bootstrap library running a search using the same sequence will yield results was used to allow display on devices of any screen size, in- nearly instantaneously. As on all other pages, results from cluding mobile devices. Several icons were included from the BLAST search page can be linked to and shared with Font Awesome and the Noun Project, and a number of other researchers using the ‘Link to these results’ link at the JavaScript libraries are used for table export and sorting. bottom of the page. For sequences above a URL length of Please see the About section of the website for detailed at- 2000 characters this link uses a sequence hash identifying tributions. the cached sequence, rather than the sequence itself. Nucleic Acids Research, 2020, Vol. 48, Database issue D513 Protein visualization (ProViz) Detailed view of infection fitness scores for a gene For UniProt proteins, a protein visualization (ProViz) au- After selecting a gene of interest, a view will open with all tomatically generated and displayed by ProViz (22). ProViz the infection fitness information available for this gene (Fig- is an interactive exploration tool for investigating the struc- ure 3). The heading of this page provides information on the tural, functional and evolutionary features of proteins and selected protein: protein and pathogen identifiers, sequence is likely to be particularly helpful for analyzing uncharac- length, gene name and UniProt identifier. terized proteins. In the table, all available experimental data are shown: Tissue name and ontology of the host organism, time after USING BACFITBASE infection, transposon insertion site on the pathogen’s main chromosome (if reported by the original study), fitness data BacFITBase consists of a text search function to find spe- during infection including the raw fitness score, normalized cific pathogenic bacterial genes, a BLAST search function fitness z-score and P-value, and the reference to the original to find bacterial genes similar to a protein or nucleic acid paper where the data were published. A brief description sequence of interest, a Browse function to quickly identify on the meaning of the raw score, normalized z-score and genes of high fitness impact during infection, and a Tutorial P-value is also available as mouse-over explanation on the section to get started quickly by following a step-by-step column headers. guide. Please note that BacFITBase relies on JavaScript, For any protein in UniProt, a ProViz is automatically therefore users will need to enable this in their web browser provided by ProViz, an interactive exploration tool for in- for full functionality. vestigating the structural, functional and evolutionary fea- tures of proteins, including Pfam domains (23) and trans- Searching for a gene or protein membrane regions. This is particularly useful for uncharac- To search for a gene or protein, users just need to type its terized proteins, which account for a large fraction of bac- name or identifier in the Search tab and press the ‘Search’ terial proteins (24). Alternatively, the protein’s FASTA se- button. Any of the following options are available: gene quence can be displayed by pressing the ‘Show protein se- symbols, gene locus identifiers, NCBI protein identifiers, quence’ button, along with a ‘Copy’ link in the top right UniProt protein accessions or a free-text search in the gene corner to copy and paste the protein’s sequence into other product’s description. The Search function also supports research tools, or into the BacFITBase BLAST Search to smart partial matches, so free-text terms can be used (e.g. search for similar proteins. Similar proteins can be searched ‘ribonuclease’). via BLAST using the ‘Find similar proteins’ button. This al- lows rapid assessment of the fitness impact of a group of similar proteins across all pathogens in BacFITBase. Searching within specific hosts and pathogens Both pathogen and/or host species can be specified in the drop-down menus on the Search page. If no search term is BLAST search given, this will result in a complete list of genes for these The BLAST Search tab provides a search by sequence simi- species, similar to the Browse view (described below). larity. When the protein of interest is not in our database, the user may search for similar proteins using BLAST sequence Search results alignment. Finding a similar protein with low z-score (and low P-value) is a strong indication that the query sequence After searching, the search results page will display a list of may be relevant for infection. any bacterial genes matching the search term and species se- To search for similar proteins in our database using lected (Figure 2). The column matched by the search term BLAST, the user can paste in the protein or coding sequence is highlighted in green (if a search term was provided). For of interest in FASTA format and then press the ‘Search’ but- each pathogen species and gene, the search results page al- ton. Both protein and coding sequences can be used, but the ready shows a preview of the lowest fitness z-score across all proper format (protein or coding sequence) must be speci- available hosts, tissues and post-infection time points, and fied in the drop-down menu next to the ‘Search’ button (as the corresponding P-value. illustrated by the examples provided on the BLAST search page). Tables on BacFITBase: sorting, downloading and linking to results BLAST results To sort any table on BacFITBase as desired, users can sim- ply click on any of the column headers. All tables can be When the BLAST alignment is ready (usually within a sec- downloaded as a comma-separated CSV file for import ond or less), a search results page will open displaying align- into spreadsheet software such as Microsoft Excel or Ap- ment performance together with a complete description ple Numbers using the ‘Download Table’ button in the top of the identified hits. This includes output columns from right corner. An appropriate, readable file name is automat- BLAST, such as the percentage of sequence identity be- ically generated. Any results pages can also be linked to and tween query and target in the successfully aligned region shared with other researchers by right-clicking and copying (‘Identity’), the total number of amino acids that were suc- the ‘Link to these results’ link at the bottom of a page or cessfully aligned between query and target (‘Aligned’), and table. the required size of a sequence database in which the current D514 Nucleic Acids Research, 2020, Vol. 48, Database issue Figure 2. Search results. This page displays a list of any bacterial genes matching the search term and the host and pathogen species that were selected. The search results page displays the gene locus and NCBI protein identifier, the UniProt accessions code and the gene symbol (Please see BacFITBase description section for further information). This preview also shows the gene product description and its length, together with the fitness z-score and the corresponding P-value. In this example, we show the case of TolB, a periplasmic component of the Tol–Pal complex that plays a central role in the maintenance of cell wall integrity. Although this complex is not essential in vitro, mutants defective in the TolB-Pal complex have a reduced infectivity. The results in BacFITBase show that, upon deletion of TolB, the infection fitness of the mutant is strongly decreased. Figure 3. Detailed view of infection fitness scores for a gene. This page displays all the infection fitness information available for a bacterial gene, along with a ProViz visualization of the protein’s sequence and structural features. The detailed view shows the fitness raw values and z-scores for each entry together with all the details of the experiment: hosts, tissues, transposon insertion sites, post-infection time points and the corresponding P-value. The reference to the original paper where the data were published is also included in the last column of the table with the corresponding PubMed link. match could be found just by chance (the ‘bit score’). The E- species and the NCBI Protein database (‘Protein’), which value is the expected number of false positive matches given provides protein sequences and information. Additionally, the size of the search database used. Matches with 100% the ‘UniProt Accession’ and ‘Gene Symbol’ columns link sequence identity in the aligned region are highlighted in out to the UniProt Knowledgebase (25), which provides green. The meaning of the Pathogen, Locus, Protein, Gene, comprehensive protein annotation. Product, P-value and Fitness z-score columns can be found in the ‘Browse Tab’ section below, or via the mouse-over in- formation symbols in the top row of any table. Downloading the entire database (the Download Tab) The entire BacFITBase database is available for download Browsing the entire database (the Browse Tab) on the Download page. Currently, BacFITBase v1 is avail- able, and will be upgraded with new data as they become The Browse tab provides an overview of all entries in the available. BacFITBase database. A pathogenic species of interest can be chosen in the selection element at the top. This table is sorted by significance and fitness z-score, which means DISCUSSION that bacterial genes with a high and significant fitness im- pact during infection are listed first. Arrows next to each Infectious diseases are caused by micro-organisms known field link out to useful external databases, including the as pathogens, which have the capacity to enter, colonize and NCBI Taxonomy database, a comprehensive taxonomic grow within a host, causing infection and damage. The use database (‘Pathogen’), the Ensembl Bacteria database (‘Lo- of antibiotics to treat bacterial infections has undoubtedly cus’), which provides genome annotation for many bacterial been one of the most important advances in healthcare, sav- Nucleic Acids Research, 2020, Vol. 48, Database issue D515 ing millions of lives since their discovery and widespread use REFERENCES (26,27). 1. Ahmed,H., Howton,T.C., Sun,Y., Weinberger,N., Belkhadir,Y. and Antibiotic development has mainly focused on the iden- Mukhtar,M.S. (2018) Network biology discovers pathogen contact tification of ‘essential’ genes and proteins in bacteria whose points in host protein-protein interactomes. Nat. Commun., 9, 1–13. inhibition is lethal under in vitro conditions (i.e. bacterial 2. Pan,A., Lahiri,C., Rajendiran,A. and Shanmugham,B. (2016) Computational analysis of protein interaction networks for infectious growth in culture). However, pathogenic bacteria do not diseases. Brief. Bioinform., 17, 517–526. grow alone but in a complex host environment. Thus, we 3. Crua Asensio,N., Munõz Giner,E., De Groot,N.S. and Torrent need to revise the definition of what ‘essential’ means so Burgas,M. (2017) Centrality in the host-pathogen interactome is that it includes the biological context of infection, i.e. which associated with pathogen fitness during infection. Nat. Commun., 8, 14092. genes are essential for the pathogen during host infection. 4. Van Opijnen,T. and Camilli,A. (2013) Transposon insertion For a given bacterium, the in vivo fitness cost of deleting sequencing: a new tool for systems-level analysis of microorganisms. a single gene is correlated with the number of interactions Nat. Rev. Microbiol., 11, 435–442. with host proteins (3). Therefore, proteins with high impact 5. Chao,M.C., Abel,S., Davis,B.M. and Waldor,M.K. (2016) The design on pathogen fitness during infection may cause extensive and analysis of transposon insertion sequencing experiments. Nat. Rev. Microbiol., 14, 119–128. rewiring of the host interactome. These observations indi- 6. Chaudhuri,R.R., Morgan,E., Peters,S.E., Pleasance,S.J., cate that infectious diseases are only properly understood Hudson,D.L., Davies,H.M., Wang,J., van Diemen,P.M., in the context of the host–pathogen interactions. Toward Buckley,A.M., Bowen,A.J. et al. (2013) Comprehensive assignment of this end, a promising approach is to systematically charac- roles for Salmonella Typhimurium genes in intestinal colonization of food-producing animals. PLoS Genet., 9, e1003456. terize proteins involved in host infection. The possibility of 7. Gawronski,J.D., Wong,S.M.S., Giannoukos,G., Ward,D. V. and accessing fitness data from disparate sources quickly and Akerley,B.J. (2009) Tracking insertion mutants within libraries by easily should accelerate the identification of new proteins deep sequencing and a genome-wide screen for Haemophilus genes involved in life-threatening infectious diseases. required in the lung. Proc. Natl. Acad. Sci. U.S.A., 106, 16422–16427. In this context, BacFITBase constitutes a valuable re- 8. Le Breton,Y., Belew,A.T., Freiberg,J.A., Sundar,G.S., Islam,E., Lieberman,J., Shirtliff,M.E., Tettelin,H., El-Sayed,N.M. and source to systematically classify bacterial proteins relevant McIver,K.S. (2017) Genome-wide discovery of novel M1T1 group A for host cell invasion and infection. BacFITBase will facil- streptococcal determinants important for fitness and virulence during itate the task of identifying target proteins and interspecies soft-tissue infection. PLoS Pathog., 13, e1006584. complexes that will help us to understand the mechanisms 9. Miller,D.P., Hutcherson,J.A., Wang,Y., Nowakowska,Z.M., Potempa,J., Yoder-Himes,D.R., Scott,D.A., Whiteley,M. and of infection, and the design of new antimicrobial molecules Lamont,R.J. (2017) Genes contributing to porphyromonas gingivalis aimed to interfere with the formation of such complexes. In fitness in abscess and epithelial cell colonization environments. Front. the next several years, we expect our database to grow con- Cell. Infect. Microbiol., 7, 378. tinuously, becoming an even more comprehensive reposi- 10. Subashchandrabose,S., Smith,S.N., Spurbeck,R.R., Kole,M.M. and tory of bacterial proteins which could be important targets Mobley,H.L.T. (2013) Genome-Wide detection of fitness genes in uropathogenic escherichia coli during systemic infection. PLoS in the fight against infectious diseases. Pathog., 9, 1–15. 11. Wang,J., Pritchard,J.R., Kreitmann,L., Montpetit,A. and Behr,M.A. (2014) Disruption of Mycobacterium avium subsp. ACKNOWLEDGEMENTS paratuberculosis-specific genes impairs in vivo fitness. BMC Genomics, 15, 415. The authors would like to thank Dr Norman E. Davey for 12. Olson,M.A., Siebach,T.W., Griffitts,J.S., Wilson,E. and his kind support in providing ProViz for protein visualiza- Erickson,D.L. (2018) Genome-wide identification of fitness factors in tion. mastitis-associated Escherichia coli. Appl. Environ. Microbiol., 84, e02190-17. 13. Eckert,S.E., Dziva,F., Chaudhuri,R.R., Langridge,G.C., Turner,D.J., Pickard,D.J., Maskell,D.J., Thomson,N.R. and Stevens,M.P. (2011) FUNDING Retrospective application of transposon-directed insertion site sequencing to a library of signature-tagged mini-Tn5Km2 mutants of Spanish Ministerio de Ciencia, Innovación y Univer- Escherichia coli O157:H7 screened in cattle. J. Bacteriol., 193, sidades [SAF2015-72518-EXP, SAF2017-82158-R, 1771–1776. RYC-2012-09999]; European Research Council [RI- 14. Fu,Y., Waldor,M.K. and Mekalanos,J.J. (2013) Tn-seq analysis of BOMYLOME 309545]; European Union’s Horizon 2020 vibrio cholerae intestinal colonization reveals a role for Research and Innovation Programme [727658, IASIS]; T6SS-mediated antibacterial activity in the host. Cell Host Microbe, 14, 652–663. Spanish Ministry of Economy and Competitiveness 15. Gao,B., Vorwerk,H., Huber,C., Lara-Tejero,M., Mohr,J., [BFU2014-55054-P, BFU2017-86970-P]; Spanish Ministry Goodman,A.L., Eisenreich,W., Galán,J.E. and Hofreuter,D. (2017) of Economy and Competitiveness; ‘Centro de Excelencia Metabolic and fitness determinants for in vitro growth and intestinal Severo Ochoa 2013–2017’; CERCA Programme of the colonization of the bacterial pathogen Campylobacter jejuni. PLoS Biol., 15, e2001390. Generalitat de Catalunya; European Union’s Horizon 2020 16. Bachman,M.A., Breen,P., Deornellas,V., Mu,Q., Zhao,L., Wu,W., Research and Innovation Programme, Marie Sklodowska- Cavalcoli,J.D. and Mobley,H.L.T. (2015) Genome-wide identification Curie Individual Fellowship [793135, ‘DeepRNA’ to B.L.]; of Klebsiella pneumoniae fitness genes during lung infection. Mbio, 6, European Society of Clinical Microbiology and Infectious e00775-15. Diseases, Research Grant 2016 (ESCMID). Funding 17. Wang,N., Ozer,E.A., Mandel,M.J. and Hauser,A.R. (2014) Genome-wide identification of Acinetobacter baumannii genes for open access charge: Spanish Ministerio de Ciencia, necessary for persistence in the lung. Mbio, 5, e01163-14. Innovación y Universidades [SAF2017-82158-R]. 18. Grant,A.J., Oshota,O., Chaudhuri,R.R., Mayho,M., Peters,S.E., Conflict of interest statement. None declared. Clare,S., Maskell,D.J. and Mastroeni,P. (2016) Genes required for the D516 Nucleic Acids Research, 2020, Vol. 48, Database issue fitness of Salmonella enterica serovar Typhimurium during infection 24. Riley,M., Abe,T., Arnaud,M.B., Berlyn,M.K.B., Blattner,F.R., of immunodeficient gp91-/- phox mice. Infect. Immun., 84, 989–997. Chaudhuri,R.R., Glasner,J.D., Horiuchi,T., Keseler,I.M., Kosuge,T. 19. Anderson,M.T., Mitchell,L.A., Zhao,L., Mobley,H.L.T. and et al. (2006) Escherichia coli K-12: A cooperatively developed Welch,R. (2017) Capsule production and glucose metabolism dictate annotation snapshot - 2005. Nucleic Acids Res., 34, 1–9. fitness during serratia marcescens bacteremia. Mbio. 8, e00740-17. 25. UniProt Consortium (2019) UniProt: a worldwide hub of protein 20. Hubbard,T.P., Chao,M.C., Abel,S., Blondel,C.J., Abel zur Wiesch,P., knowledge. Nucleic Acids Res., 47, D506–D515. Zhou,X., Davis,B.M. and Waldor,M.K. (2016) Genetic analysis of 26. Cassini,A., Högberg,L.D., Plachouras,D., Quattrocchi,A., Hoxha,A., Vibrio parahaemolyticus intestinal colonization. Proc. Natl. Acad. Simonsen,G.S., Colomb-Cotinat,M., Kretzschmar,M.E., Sci. U.S.A., 113, 6283–6288. Devleesschauwer,B., Cecchini,M. et al. (2019) Attributable deaths 21. Camacho,C., Coulouris,G., Avagyan,V., Ma,N., Papadopoulos,J., and disability-adjusted life-years caused by infections with Bealer,K. and Madden,T.L. (2009) BLAST+: architecture and antibiotic-resistant bacteria in the EU and the European Economic applications. BMC Bioinformatics, 10, 421. Area in 2015: a population-level modelling analysis. Lancet Infect. 22. Jehl,P., Manguy,J., Shields,D.C., Higgins,D.G. and Davey,N.E. Dis., 19, 56–66. (2016) ProViz-a web-based visualization tool to investigate the 27. Collignon,P., Beggs,J.J., Walsh,T.R., Gandra,S. and Laxminarayan,R. functional and evolutionary features of protein sequences. Nucleic (2018) Anthropological and socioeconomic factors contributing to Acids Res., 44, W11–W15. global antimicrobial resistance: a univariate and multivariable 23. El-Gebali,S., Mistry,J., Bateman,A., Eddy,S.R., Luciani,A., analysis. Lancet Planet. Heal., 2, e398–e405. Potter,S.C., Qureshi,M., Richardson,L.J., Salazar,G.A., Smart,A. et al. (2019) The Pfam protein families database in 2019. Nucleic Acids Res., 47, D427–D432.