Tag: Metagenomics

Since F1000Research uses a somewhat different publication scheme than most journals, I still haven’t understood if this paper is formally published after peer review, but I start to assume it is. There have been very little changes since the last version, so hence I will be lazy and basically repost what I wrote in April when the first version (the “preprint”) was posted online. The paper (1) is the result of a workshop arranged by the JRC in Italy in 2017. It describes various challenges arising from the process of designing a benchmark strategy for bioinformatics pipelines in the identification of antimicrobial resistance genes in next generation sequencing data.

The paper discusses issues about the benchmarking datasets used, testing samples, evaluation criteria for the performance of different tools, and how the benchmarking dataset should be created and distributed. Specially, we address the following questions:

• How should a benchmark strategy handle the current and expanding universe of NGS platforms?
• What should be the quality profile (in terms of read length, error rate, etc.) of in silico reference materials?
• Should different sets of reference materials be produced for each platform? In that case, how to ensure no bias is introduced in the process?
• Should in silico reference material be composed of the output of real experiments, or simulated read sets? If a combination is used, what is the optimal ratio?
• How is it possible to ensure that the simulated output has been simulated “correctly”?
• For real experiment datasets, how to avoid the presence of sensitive information?
• Regarding the quality metrics in the benchmark datasets (e.g. error rate, read quality), should these values be fixed for all datasets, or fall within specific ranges? How wide can/should these ranges be?
• How should the benchmark manage the different mechanisms by which bacteria acquire resistance?
• What is the set of resistance genes/mechanisms that need to be included in the benchmark? How should this set be agreed upon?
• Should datasets representing different sample types (e.g. isolated clones, environmental samples) be included in the same benchmark?
• Is a correct representation of different bacterial species (host genomes) important?
• How can the “true” value of the samples, against which the pipelines will be evaluated, be guaranteed?
• What is needed to demonstrate that the original sample has been correctly characterised, in case real experiments are used?
• How should the target performance thresholds (e.g. specificity, sensitivity, accuracy) for the benchmark suite be set?
• What is the impact of these performance thresholds on the required size of the sample set?
• How can the benchmark stay relevant when new resistance mechanisms are regularly characterized?
• How is the continued quality of the benchmark dataset ensured?
• Who should generate the benchmark resource?
• How can the benchmark resource be efficiently shared?

Of course, we have not answered all these questions, but I think we have come down to a decent description of the problems, which we see as an important foundation for solving these issues and implementing the benchmarking standard. Some of these issues were tackled in our review paper from last year on using metagenomics to study resistance genes in microbial communities (2). The paper also somewhat connects to the database curation paper we published in 2016 (3), although this time the strategies deal with the testing datasets rather than the actual databases. The paper is the first outcome of the workshop arranged by the JRC on “Next-generation sequencing technologies and antimicrobial resistance” held October 4-5 2017 in Ispra, Italy. You can find the paper here (it’s open access).

On another note, the new paper describing the UNITE database (4) has now got a formal issue assigned to it, as has the paper on tandem repeat barcoding in fungi published in Molecular Ecology Resources last year (5).

References and notes

1. Angers-Loustau A, Petrillo M, Bengtsson-Palme J, Berendonk T, Blais B, Chan KG, Coque TM, Hammer P, Heß S, Kagkli DM, Krumbiegel C, Lanza VF, Madec J-Y, Naas T, O’Grady J, Paracchini V, Rossen JWA, Ruppé E, Vamathevan J, Venturi V, Van den Eede G: The challenges of designing a benchmark strategy for bioinformatics pipelines in the identification of antimicrobial resistance determinants using next generation sequencing technologies. F1000Research, 7, 459 (2018). doi: 10.12688/f1000research.14509.1
2. Bengtsson-Palme J, Larsson DGJ, Kristiansson E: Using metagenomics to investigate human and environmental resistomes. Journal of Antimicrobial Chemotherapy, 72, 2690–2703 (2017). doi: 10.1093/jac/dkx199
3. Bengtsson-Palme J, Boulund F, Edström R, Feizi A, Johnning A, Jonsson VA, Karlsson FH, Pal C, Pereira MB, Rehammar A, Sánchez J, Sanli K, Thorell K: Strategies to improve usability and preserve accuracy in biological sequence databases. Proteomics, 16, 18, 2454–2460 (2016). doi: 10.1002/pmic.201600034
4. Nilsson RH, Larsson K-H, Taylor AFS, Bengtsson-Palme J, Jeppesen TS, Schigel D, Kennedy P, Picard K, Glöckner FO, Tedersoo L, Saar I, Kõljalg U, Abarenkov K: The UNITE database for molecular identification of fungi: handling dark taxa and parallel taxonomic classifications. Nucleic Acids Research, 47, D1, D259–D264 (2019). doi: 10.1093/nar/gky1022
5. Wurzbacher C, Larsson E, Bengtsson-Palme J, Van den Wyngaert S, Svantesson S, Kristiansson E, Kagami M, Nilsson RH: Introducing ribosomal tandem repeat barcoding for fungi. Molecular Ecology Resources, 19, 1, 118–127 (2019). doi: 10.1111/1755-0998.12944

Let me get straight to something somewhat besides the point here: summer students can achieve amazing things! One such student I had the pleasure to work with this summer is Shruthi Magesh, and a preprint based on work she did with me at the Wisconsin Institute for Discovery this summer just got published on bioRxiv (1). The preprint describes a software tool called Mumame, which uses database information on mutations in DNA or protein sequences to search metagenomic datasets and quantifies the relative proportion of resistance mutations over wild type sequences.

In the preprint (1), we first of all show that Mumame works on amplicon data where we already knew the true outcome (2). Second, we show that we can detect differences in mutation frequencies in controlled experiments (2,3). Lastly, we use the tool to gain some further information about resistance patterns in sediments from polluted environments in India (4,5). Together these analyses show that one of the most central aspects for Mumame to be able to find mutations is having a very high number of sequenced reads in all libraries (preferably more than 50 million per library), because these mutations are generally rare – even in polluted environments and microcosms exposed to antibiotics. We expect Mumame to be a useful addition to metagenomic studies of e.g. antibiotic resistance, and to increase the detail by which metagenomes can be screened for phenotypically important differences.

While I did write the code for the software (with a lot of input from Viktor Jonsson, who also is a coauthor on the preprint, on the statistical analysis), Shruthi did the software testing and evaluations, and the paper would not have been possible hadn’t she wanted a bioinformatic summer project related to metagenomics, aside from her laboratory work. The resulting preprint is available from bioRxiv and the Mumame software is freely available from this site.

References

1. Magesh S, Jonsson V, Bengtsson-Palme J: Quantifying point-mutations in metagenomic data. bioRxiv, 438572 (2018). doi: 10.1101/438572 [Link]
2. Kraupner N, Ebmeyer S, Bengtsson-Palme J, Fick J, Kristiansson E, Flach C-F, Larsson DGJ: Selective concentration for ciprofloxacin in Escherichia coli grown in complex aquatic bacterial biofilms. Environment International, 116, 255–268 (2018). doi: 10.1016/j.envint.2018.04.029 [Paper link]
3. Lundström S, Östman M, Bengtsson-Palme J, Rutgersson C, Thoudal M, Sircar T, Blanck H, Eriksson KM, Tysklind M, Flach C-F, Larsson DGJ: Minimal selective concentrations of tetracycline in complex aquatic bacterial biofilms. Science of the Total Environment, 553, 587–595 (2016). doi: 10.1016/j.scitotenv.2016.02.103 [Paper link]
4. Bengtsson-Palme J, Boulund F, Fick J, Kristiansson E, Larsson DGJ: Shotgun metagenomics reveals a wide array of antibiotic resistance genes and mobile elements in a polluted lake in India. Frontiers in Microbiology, 5, 648 (2014). doi: 10.3389/fmicb.2014.00648 [Paper link]
5. Kristiansson E, Fick J, Janzon A, Grabic R, Rutgersson C, Weijdegård B, Söderström H, Larsson DGJ: Pyrosequencing of antibiotic-contaminated river sediments reveals high levels of resistance and gene transfer elements. PLoS ONE, Volume 6, e17038 (2011). doi:10.1371/journal.pone.0017038.

This week, a paper by my former roommate Katariina Pärnänen was published by Nature Communications. In the paper (1), we use shotgun metagenomics to show that infants carry more resistant bacteria in their gut than adults do, irrespective of whether they themselves have been treated with antibiotics or not. We also found that the antibiotic resistance gene and mobile genetic element profiles of infant feces are more similar to those of their own mothers than to those of unrelated mothers. This is suggestive of a pathway of transmission of resistance genes from the mothers, and importantly we find that the mobile genetic elements in breastmilk are shared with those of the infant feces, despite vast differences in their microbiota composition. Finally, we find that termination of breastfeeding and intrapartum antibiotic prophylaxis of mothers are associated with higher abundances of specific ARGs in the infant gut. Our results suggest that infants inherit the legacy of past antibiotic consumption of their mothers via transmission of genes, but that the taxonomic composition of the microbiota still strongly dictates the overall load of resistance genes.

I am not going to dwell in to details of the study here, but I instead encourage you to read the paper (hey, it’s open access!) or the excellent popular summary that Katariina has already written. Finally, I want to emphasize the great work Katariina has put into this (I would know, since I shared room with her) and congratulate her on her own little infant!

Reference

1. Pärnänen K, Karkman A, Hultman J, Lyra C, Bengtsson-Palme J, Larsson DGJ, Rautava S, Isolauri E, Salminen S, Kumar H, Satokari R, Virta M: Maternal gut and breast milk microbiota affect infant gut antibiotic resistome and mobile genetic elements. Nature Communications, 9, 3891 (2018). doi: 10.1038/s41467-018-06393-w [Paper link]

I’m really late at this ball for a number of reasons, but last week Nature published our paper on the structure and function of the global topsoil microbiome (1). This paper has a long story, but in short I got contacted by Mohammad Bahram (the first author) about two years ago about a project using metagenomic sequencing to look at a lot of soil samples for patterns of antibiotic resistance gene abundances and diversity. The project had made the interesting discovery that resistance gene abundances were linked to the ratio of fungi and bacteria (so that more fungi was linked to more resistance genes). During the following year, we together worked on deciphering these discoveries, which are now published in Nature. The paper also deals with the taxonomic patterns linked to geography (1), but as evident from the above, my main contribution here has been on the antibiotic resistance side.

In short, we find that:

• Bacterial diversity is highest in temperate habitats, and lower both closer to the equator and the poles
• For bacteria, the diversity of biological functions follows the same pattern, but for fungi, the functional diversity is higher closer to the poles and the equator
• Higher abundance of fungi is linked to higher abundance and diversity of antibiotic resistance genes. Specifically, this is related to known antibiotic producing fungal lineages, such as Penicillium and Oidiodendron. There also seems to be a link between the Actinobacteria, encompassing the antibiotic-producing bacterial genus of Streptomyces and higher resistance gene diversity.
• Similar relationships between the fungus-like Oomycetes and resistance genes was also found in ocean samples from the Tara Oceans project (2)

The results of this study indicate that both environmental filtering and niche differentiation determine soil microbial composition, and that the role of dispersal limitation is minor at this scale. Soil pH and precipitation seems to be the strongest drivers of community composition. Furthermore, we interpret our data to reveal that inter-kingdom antagonism is important in structuring microbial communities. This speaks against the notion put forward that antibiotic resistance genes might not have a resistance function in natural settings (3). That said, the most likely explanation here is probably a bit of both warfare and repurposing of genes. Soil seems to be the largest untapped source of resistance genes for human pathogens (4), and the finding that natural antagonism may be driving resistance gene diversification and enrichment may be important for future management of environmental antibiotic resistance (5,6).

It was really great to work with Mohammad and his team, and I sure hope that we will collaborate again in the future. The entire paper can be found in the issue of Nature coming out this week, and is already online at Nature’s website.

References

1. Bahram M°, Hildebrand F°, Forslund SK, Anderson JL, Soudzilovskaia NA, Bodegom PM, Bengtsson-Palme J, Anslan S, Coelho LP, Harend H, Huerta-Cepas J, Medema MH, Maltz MR, Mundra S, Olsson PA, Pent M, Põlme S, Sunagawa S, Ryberg M, Tedersoo L, Bork P: Structure and function of the global topsoil microbiome. Nature, 560, 233–237 (2018). doi: 10.1038/s41586-018-0386-6
2. Sunagawa S et al. Structure and function of the global ocean microbiome. Science 348, 6237, 1261359 (2015). doi: 10.1126/science.1261359
3. Aminov RI: The role of antibiotics and antibiotic resistance in nature. Environmental Microbiology, 11, 12, 2970-2988 (2009). doi: 10.1111/j.1462-2920.2009.01972.x
4. Bengtsson-Palme J: The diversity of uncharacterized antibiotic resistance genes can be predicted from known gene variants – but not always. Microbiome, 6, 125 (2018). doi: 10.1186/s40168-018-0508-2
5. Bengtsson-Palme J, Kristiansson E, Larsson DGJ: Environmental factors influencing the development and spread of antibiotic resistance. FEMS Microbiology Reviews, 42, 1, 68–80 (2018). doi: 10.1093/femsre/fux053
6. Larsson DGJ, Andremont A, Bengtsson-Palme J, Brandt KK, de Roda Husman AM, Fagerstedt P, Fick J, Flach C-F, Gaze WH, Kuroda M, Kvint K, Laxminarayan R, Manaia CM, Nielsen KM, Ploy M-C, Segovia C, Simonet P, Smalla K, Snape J, Topp E, van Hengel A, Verner-Jeffreys DW, Virta MPJ, Wellington EM, Wernersson A-S: Critical knowledge gaps and research needs related to the environmental dimensions of antibiotic resistance. Environment International, 117, 132–138 (2018).

I have just published a popular-science-for-scientists type of post at the Nature Microbiology Community about my recent paper published in Microbiome. I personally think that it might be worth a read, so feel free to head over here and read it!

Over the weekend, Microbiome put online my most recent paper (1) – a project which started as an idea I got when I finished up my PhD thesis in 2016. One of my main points in the thesis (2), which was also made again on our recent review on environmental factors influencing resistance development (3), is that the greatest risks associated with antibiotic resistance in the environment may not be the resistance genes already circulating in pathogens (which are relatively easily quantified), but the ones associated with recruitment of novel resistance genes from bacteria in the environment (2-4). The latter genes are, however, impossible to quantify due to the fact that they are unknown. But what if we could use knowledge of the diversity and abundance of known resistance genes to estimate the same properties of the yet uncharacterized resistome? That would be a great advantage in e.g. ranking of risk environments, as then some property that is easily monitored can be used to inform risk management of both known and unknown resistance factors.

This just published paper explores this possibility, by quantifying the abundance and diversity of resistance genes in 1109 metagenomes from various environments (1). I have taken two different approaches. First, I took out smaller subsets of genes from the reference database (in this case Resqu, a database of antibiotic resistance genes with verified resistance functions, detected on mobile genetic elements), and used those subsets to estimate resistome diversity and abundance in the 1109 metagenomes. Then these predictions were compared to the results of the entire database. I then, in a second step, investigated if these predictions could be extended to a set of truly novel resistance genes, i.e. the resistance genes present in the FARME database, collecting data from functional metagenomics inserts (5,6).

The results show that generally the diversity and abundance of known antibiotic resistance genes can be used to predict the same properties of undescribed resistance genes (see figure above). However, the extent of this predictability is, importantly, dependent on the type of environment investigated. The study also shows that carefully selected small sets of resistance genes can describe total resistance gene diversity remarkably well. This means that knowledge gained from large-scale quantifications of known resistance genes can be utilized as a proxy for unknown resistance factors. This is important for current and proposed monitoring efforts for environmental antibiotic resistance (7-11) and has implications for the design of risk ranking strategies and the choices of measures and methods for describing resistance gene abundance and diversity in the environment.

The study also investigated which diversity measures were best suited to estimate total diversity. Surprisingly, some diversity measures described the total diversity of resistance genes remarkably bad. Most prominently, the Simpson diversity index consistently showed poor performance, and while the Shannon index performed relatively better, there is still no reason to select the Shannon index over normalized (rarefied) richness of resistance genes. The ACE estimator fluctuated substantially compared to the other diversity measures, while the Chao1 estimator more consistently showed performance very similar to richness. Therefore, either richness or the Chao1 estimator should be used for ranking resistance gene diversity, while the Shannon, Simpson, and ACE measures should be avoided.

Importantly, this study implies that the recruitment of novel antibiotic resistance genes from the environment to human pathogens is essentially random. Therefore, when ranking risks associated with antibiotic resistance in environmental settings, the knowledge gained from large-scale quantification of known resistance genes can be utilized as a proxy for the unknown resistance factors (although this proxy is not perfect). Thus, high-risk environments for resistance development and dissemination would, for example, be aquaculture, animal husbandry, discharges from antibiotic manufacturing, and untreated sewage (3,8,12-15). Further attention should probably be paid to antibiotic contaminated soils, as this study points to soils as a vast source of resistance genes not yet encountered in human pathogens. This has also been suggested previously by others (16-19). The results of this study can be used to guide monitoring efforts for environmental antibiotic resistance, to design risk ranking strategies, and to choose appropriate measures and methods for describing resistance gene abundance and diversity in the environment. The entire open access paper is available here.

References

1. Bengtsson-Palme J: The diversity of uncharacterized antibiotic resistance genes can be predicted from known gene variants – but not always. Microbiome, 6, 125 (2018). doi: 10.1186/s40168-018-0508-2
2. Bengtsson-Palme J: Antibiotic resistance in the environment: a contribution from metagenomic studies. Doctoral thesis (medicine), Department of Infectious Diseases, Institute of Biomedicine, Sahlgrenska Academy, University of Gothenburg, 2016. [Link]
3. Bengtsson-Palme J, Kristiansson E, Larsson DGJ: Environmental factors influencing the development and spread of antibiotic resistance. FEMS Microbiology Reviews, 42, 1, 68–80 (2018). doi: 10.1093/femsre/fux053
4. Bengtsson-Palme J, Larsson DGJ: Antibiotic resistance genes in the environment: prioritizing risks. Nature Reviews Microbiology, 13, 369 (2015). doi: 10.1038/nrmicro3399-c1
5. Wallace JC, Port JA, Smith MN, Faustian EM: FARME DB: a functional antibiotic resistance element database. Database, 2017, baw165 (2017).
6. Handelsman J, Rondon MR, Brady SF, Clardy J, Goodman RM: Molecular biological access to the chemistry of unknown soil microbes: a new frontier for natural products. Chemical Biology, 5, R245–249 (1998).
7. Berendonk TU, Manaia CM, Merlin C, Fatta-Kassinos D, Cytryn E, Walsh F, et al.: Tackling antibiotic resistance: the environmental framework. Nature Reviews Microbiology, 13, 310–317 (2015).
8. Pruden A, Larsson DGJ, Amézquita A, Collignon P, Brandt KK, Graham DW, et al.: Management options for reducing the release of antibiotics and antibiotic resistance genes to the environment. Environmental Health Perspectives, 121, 878–885 (2013).
9. Review on Antimicrobial Resistance: Antimicrobials in agriculture and the environment: reducing unnecessary use and waste. O’Neill J, ed. London: Wellcome Trust & HM Government (2015).
10. Angers-Loustau A, Petrillo M, Bengtsson-Palme J, Berendonk T, Blais B, Chan KG, Coque TM, Hammer P, Heß S, Kagkli DM, Krumbiegel C, Lanza VF, Madec J-Y, Naas T, O’Grady J, Paracchini V, Rossen JWA, Ruppé E, Vamathevan J, Venturi V, Van den Eede G: The challenges of designing a benchmark strategy for bioinformatics pipelines in the identification of antimicrobial resistance determinants using next generation sequencing technologies. F1000Research, 7, 459 (2018). doi: 10.12688/f1000research.14509.1
11. Larsson DGJ, Andremont A, Bengtsson-Palme J, Brandt KK, de Roda Husman AM, Fagerstedt P, Fick J, Flach C-F, Gaze WH, Kuroda M, Kvint K, Laxminarayan R, Manaia CM, Nielsen KM, Ploy M-C, Segovia C, Simonet P, Smalla K, Snape J, Topp E, van Hengel A, Verner-Jeffreys DW, Virta MPJ, Wellington EM, Wernersson A-S: Critical knowledge gaps and research needs related to the environmental dimensions of antibiotic resistance. Environment International, 117, 132–138 (2018). doi: 10.1016/j.envint.2018.04.041
12. Allen HK, Donato J, Wang HH, Cloud-Hansen KA, Davies J, Handelsman J: Call of the wild: antibiotic resistance genes in natural environments. Nature Reviews Microbiology, 8, 251–259 (2010).
13. Graham DW, Collignon P, Davies J, Larsson DGJ, Snape J: Underappreciated role of regionally poor water quality on globally increasing antibiotic resistance. Environmental Science & Technology, 48,11746–11747 (2014).
14. Larsson DGJ: Pollution from drug manufacturing: review and perspectives. Philosophical Transactions of the Royal Society of London, Series B Biological Sciences, 369, 20130571 (2014).
15. Cabello FC, Godfrey HP, Buschmann AH, Dölz HJ: Aquaculture as yet another environmental gateway to the development and globalisation of antimicrobial resistance. Lancet Infectious Diseases, 16, e127–133 (2016).
16. Forsberg KJ, Reyes A, Wang B, Selleck EM, Sommer MOA, Dantas G: The shared antibiotic resistome of soil bacteria and human pathogens. Science, 337, 1107–1111 (2012).
17. Allen HK, Moe LA, Rodbumrer J, Gaarder A, Handelsman J: Functional metagenomics reveals diverse beta-lactamases in a remote Alaskan soil. ISME Journal, 3, 243–251 (2009).
18. Riesenfeld CS, Goodman RM, Handelsman J: Uncultured soil bacteria are a reservoir of new antibiotic resistance genes. Environmental Microbiology, 6, 981–989 (2004).
19. McGarvey KM, Queitsch K, Fields S: Wide variation in antibiotic resistance proteins identified by functional metagenomic screening of a soil DNA library. Applied and Environmental Microbiology, 78, 1708–1714 (2012).

One of the questions I have received regarding Metaxa2 is if it is possible to use it on other DNA barcodes. My answer has been “technically, yes, but it is a very cumbersome process of creating a custom database for every additional barcode”. Not anymore, the newly introduced Metaxa2 Database Builder makes this process automatic, with the user just supplying a FASTA file of sequences from the region in question and a file containing the taxonomy information for the sequences (in GenBank, NSD XML, Metaxa2 or SILVA-style formats). The preprint (1) has been out for some time, but today Bioinformatics published the paper describing the software (2).

The paper not only details how the database builder works, but also shows that it is working on a number of different barcoding regions, albeit with different results in terms of accuracy. Still, even with seemingly high misclassification rates for some DNA barcodes, the software performs better than a simple BLAST-based taxonomic assignment (76.5% vs. 41.4% correct classifications for matK, and 76.2% vs. 45.1% for tnrL). The database builder has already found use in building a COI database for anthropods (3), and we envision a range of uses in the near future.

As the paper is now published, I have also moved the Metaxa2 software (4) from beta-status to a full-worthy version 2.2 update. Hopefully, this release should be bug free, but my experience is that when the community gets their hands of the software they tend to discover things our team has missed. I would like to thank the entire team working on this, particularly Rodney Richardson (who initiated this entire thing) and Henrik Nilsson. The software can be downloaded here. Happy barcoding!

References

1. Bengtsson-Palme J, Richardson RT, Meola M, Wurzbacher C, Tremblay ED, Thorell K, Kanger K, Eriksson KM, Bilodeau GJ, Johnson RM, Hartmann M, Nilsson RH: Taxonomic identification from metagenomic or metabarcoding data using any genetic marker. bioRxiv 253377 (2018). doi: 10.1101/253377 [Link]
2. Bengtsson-Palme J, Richardson RT, Meola M, Wurzbacher C, Tremblay ED, Thorell K, Kanger K, Eriksson KM, Bilodeau GJ, Johnson RM, Hartmann M, Nilsson RH: Metaxa2 Database Builder: Enabling taxonomic identification from metagenomic and metabarcoding data using any genetic marker. Bioinformatics, advance article (2018). doi: 10.1093/bioinformatics/bty482 [Paper link]
3. Richardson RT, Bengtsson-Palme J, Gardiner MM, Johnson RM: A reference cytochrome c oxidase subunit I database curated for hierarchical classification of arthropod metabarcoding data. PeerJ Preprints, 6, e26662v1 (2018). doi: 10.7287/peerj.preprints.26662v1 [Link]
4. Bengtsson-Palme J, Hartmann M, Eriksson KM, Pal C, Thorell K, Larsson DGJ, Nilsson RH: Metaxa2: Improved identification and taxonomic classification of small and large subunit rRNA in metagenomic data. Molecular Ecology Resources, 15, 6, 1403–1414 (2015). doi: 10.1111/1755-0998.12399 [Paper link]

My colleagues in Gothenburg have published a new paper in Environment International, in which I was involved in the bioinformatics analyses. In the paper, for which Nadine Kraupner did the lion’s share of the work, we establish minimal selective concentrations (MSCs) for resistance to the antibiotic ciprofloxacin in Escherichia coli grown in complex microbial communities (1). We also determine the community responses at the taxonomic and resistance gene levels. Nadine has made use of Sara Lundström’s aquarium system (2) to grow biofilms in the exposure of sublethal levels of antibiotics. Using the system, we find that 1 μg/L ciprofloxacin selects for the resistance gene qnrD, while 10 μg/L ciprofloxacin is required to detect changes of phenotypic resistance. In short, the different endpoints studied (and their corresponding MSCs) were:

• CFU counts from test tubes, grown on R2A plates with 2 mg/L ciprofloxain – MSC = 5 μg/L
• CFU counts from aquaria, grown on R2A plates with 0.25 or 2 mg/L ciprofloxain – MSC = 10 μg/L
• Chromosomal resistance mutations – MSC ~ 10 μg/L
• Increased resistance gene abundances, metagenomics – MSC range: 1 μg/L
• Changes to taxonomic diversity – 1 µg/L
• Changes to taxonomic community composition – MSC ~ 1-10 μg/L

We have previously reported a predicted no-effect concentration for resistance of 0.064 µg/L for ciprofloxacin (3), which corresponds fairly well with the MSCs determined experimentally here, being around a factor of ten off. However, we cannot exclude that in other experimental systems, the selective effects of ciprofloxacin could be even lower and thus the predicted PNEC may still be relevant. The selective concentrations we report for ciprofloxacin are close to those that have been reported in sewage treatment plants (3-5), suggesting the possibility for weak selection of resistance. Several recent reports have underscored the need to populate the this far conceptual models for resistance development in the environment with actual numbers (6-10). Determining selective concentrations for different antibiotics in actual community settings is an important step on the road towards building accurate quantitative models for resistance emergence and propagation.

References

1. Kraupner N, Ebmeyer S, Bengtsson-Palme J, Fick J, Kristiansson E, Flach C-F, Larsson DGJ: Selective concentration for ciprofloxacin in Escherichia coli grown in complex aquatic bacterial biofilms. Environment International, 116, 255–268 (2018). doi: 10.1016/j.envint.2018.04.029 [Paper link]
2. Lundström SV, Östman M, Bengtsson-Palme J, Rutgersson C, Thoudal M, Sircar T, Blanck H, Eriksson KM, Tysklind M, Flach C-F, Larsson DGJ: Minimal selective concentrations of tetracycline in complex aquatic bacterial biofilms. Science of the Total Environment, 553, 587–595 (2016). doi: 10.1016/j.scitotenv.2016.02.103 [Paper link]
3. Bengtsson-Palme J, Larsson DGJ: Concentrations of antibiotics predicted to select for resistant bacteria: Proposed limits for environmental regulation. Environment International, 86, 140-149 (2016). doi: 10.1016/j.envint.2015.10.015
4. Michael I, Rizzo L, McArdell CS, Manaia CM, Merlin C, Schwartz T, Dagot C, Fatta-Kassinos D: Urban wastewater treatment plants as hotspots for the release of antibiotics in the environment: a review. Water Research, 47, 957–995 (2013). doi:10.1016/j.watres.2012.11.027
5. Bengtsson-Palme J, Hammarén R, Pal C, Östman M, Björlenius B, Flach C-F, Kristiansson E, Fick J, Tysklind M, Larsson DGJ: Elucidating selection processes for antibiotic resistance in sewage treatment plants using metagenomics. Science of the Total Environment, 572, 697–712 (2016). doi: 10.1016/j.scitotenv.2016.06.228
6. Ågerstrand M, Berg C, Björlenius B, Breitholtz M, Brunstrom B, Fick J, Gunnarsson L, Larsson DGJ, Sumpter JP, Tysklind M, Rudén C: Improving environmental risk assessment of human pharmaceuticals. Environmental Science and Technology (2015). doi:10.1021/acs.est.5b00302
7. Bengtsson-Palme J, Kristiansson E, Larsson DGJ: Environmental factors influencing the development and spread of antibiotic resistance. FEMS Microbiology Reviews, 42, 1, 68–80 (2018). doi: 10.1093/femsre/fux053
8. Joint Programming Initiative on Antimicrobial Resistance: JPIAMR Workshop on Environmental Dimensions of AMR: Summary and recommendations. JPIAMR (2017). [Link]
9. Angers A, Petrillo P, Patak, A, Querci M, Van den Eede G: The Role and Implementation of Next-Generation Sequencing Technologies in the Coordinated Action Plan against Antimicrobial Resistance. JRC Conference and Workshop Report, EUR 28619 (2017). doi: 10.2760/745099
10. Larsson DGJ, Andremont A, Bengtsson-Palme J, Brandt KK, de Roda Husman AM, Fagerstedt P, Fick J, Flach C-F, Gaze WH, Kuroda M, Kvint K, Laxminarayan R, Manaia CM, Nielsen KM, Ploy M-C, Segovia C, Simonet P, Smalla K, Snape J, Topp E, van Hengel A, Verner-Jeffreys DW, Virta MPJ, Wellington EM, Wernersson A-S: Critical knowledge gaps and research needs related to the environmental dimensions of antibiotic resistance. Environment International, in press (2018). doi: 10.1016/j.envint.2018.04.041

This weekend, F1000Research put online the non-peer-reviewed version of the paper resulting from a workshop arranged by the JRC in Italy last year (1). (I will refer to this as a preprint, but at F1000Research the line is quite blurry between preprint and published paper.) The paper describes various challenges arising from the process of designing a benchmark strategy for bioinformatics pipelines (2) in the identification of antimicrobial resistance genes in next generation sequencing data.

The paper discusses issues about the benchmarking datasets used, testing samples, evaluation criteria for the performance of different tools, and how the benchmarking dataset should be created and distributed. Specially, we address the following questions:

• How should a benchmark strategy handle the current and expanding universe of NGS platforms?
• What should be the quality profile (in terms of read length, error rate, etc.) of in silico reference materials?
• Should different sets of reference materials be produced for each platform? In that case, how to ensure no bias is introduced in the process?
• Should in silico reference material be composed of the output of real experiments, or simulated read sets? If a combination is used, what is the optimal ratio?
• How is it possible to ensure that the simulated output has been simulated “correctly”?
• For real experiment datasets, how to avoid the presence of sensitive information?
• Regarding the quality metrics in the benchmark datasets (e.g. error rate, read quality), should these values be fixed for all datasets, or fall within specific ranges? How wide can/should these ranges be?
• How should the benchmark manage the different mechanisms by which bacteria acquire resistance?
• What is the set of resistance genes/mechanisms that need to be included in the benchmark? How should this set be agreed upon?
• Should datasets representing different sample types (e.g. isolated clones, environmental samples) be included in the same benchmark?
• Is a correct representation of different bacterial species (host genomes) important?
• How can the “true” value of the samples, against which the pipelines will be evaluated, be guaranteed?
• What is needed to demonstrate that the original sample has been correctly characterised, in case real experiments are used?
• How should the target performance thresholds (e.g. specificity, sensitivity, accuracy) for the benchmark suite be set?
• What is the impact of these performance thresholds on the required size of the sample set?
• How can the benchmark stay relevant when new resistance mechanisms are regularly characterized?
• How is the continued quality of the benchmark dataset ensured?
• Who should generate the benchmark resource?
• How can the benchmark resource be efficiently shared?

Of course, we have not answered all these questions, but I think we have come down to a decent description of the problems, which we see as an important foundation for solving these issues and implementing the benchmarking standard. Some of these issues were tackled in our review paper from last year on using metagenomics to study resistance genes in microbial communities (3). The paper also somewhat connects to the database curation paper we published in 2016 (4), although this time the strategies deal with the testing datasets rather than the actual databases. The paper is the first outcome of the workshop arranged by the JRC on “Next-generation sequencing technologies and antimicrobial resistance” held October 4-5 last year in Ispra, Italy. You can find the paper here (it’s open access).

References and notes

1. Angers-Loustau A, Petrillo M, Bengtsson-Palme J, Berendonk T, Blais B, Chan KG, Coque TM, Hammer P, Heß S, Kagkli DM, Krumbiegel C, Lanza VF, Madec J-Y, Naas T, O’Grady J, Paracchini V, Rossen JWA, Ruppé E, Vamathevan J, Venturi V, Van den Eede G: The challenges of designing a benchmark strategy for bioinformatics pipelines in the identification of antimicrobial resistance determinants using next generation sequencing technologies. F1000Research, 7, 459 (2018). doi: 10.12688/f1000research.14509.1
2. You may remember that I hate the term “pipeline” for bioinformatics protocols. I would have preferred if it was called workflows or similar, but the term “pipeline” has taken hold and I guess this is a battle where I have essentially lost. The bioinformatics workflows will be known as pipelines, for better and worse.
3. Bengtsson-Palme J, Larsson DGJ, Kristiansson E: Using metagenomics to investigate human and environmental resistomes. Journal of Antimicrobial Chemotherapy, 72, 2690–2703 (2017). doi: 10.1093/jac/dkx199
4. Bengtsson-Palme J, Boulund F, Edström R, Feizi A, Johnning A, Jonsson VA, Karlsson FH, Pal C, Pereira MB, Rehammar A, Sánchez J, Sanli K, Thorell K: Strategies to improve usability and preserve accuracy in biological sequence databases. Proteomics, 16, 18, 2454–2460 (2016). doi: 10.1002/pmic.201600034

I am very happy to announce that a first public beta version of Metaxa2 version 2.2 has been released today! This new version brings two big and a number of small improvements to the Metaxa2 software (1). The first major addition is the introduction of the Metaxa2 Database Builder, which allows the user to create custom databases for virtually any genetic barcoding region. The second addition, which is related to the first, is that the classifier has been rewritten to have a more solid mathematical foundation. I have been promising that these updates were coming “soon” for one and a half years, but finally the end-product is good enough to see some real world testing. Bear in mind though that this is still a beta version that could contain obscure bugs. Here follows a list of new features (with further elaboration on a few below):

• The Metaxa2 Database Builder
• Support for additional barcoding genes, virtually any genetic region can now be used for taxonomic classification in Metaxa2
• The Metaxa2 database repository, which can be accessed through the new metaxa2_install_database tool
• Improved classification scoring model for better clarity and sensitivity
• A bundled COI database for athropods, showing off the capabilities of the database builder
• Support for compressed input files (gzip, zip, bzip, dsrc)
• Support for auto-detection of database locations
• Added output of probable taxonomic origin for sequences with reliability scores at each rank, made possible by the updated classifier
• Added the -x option for running only the extraction without the classification step
• Improved memory handling for very large rRNA datasets in the classifier (millions of sequences)
• This update also fixes a bug in the metaxa2_rf tool that could cause bias in very skewed datasets with small numbers of taxa

The new version of Metaxa2 can be downloaded here, and for those interested I will spend the rest of this post outlining the Metaxa2 Database Builder. The information below is also available in a slightly extended version in the software manual.

The major enhancement in Metaxa2 version 2.2 is the ability to use custom databases for classification. This means that the user can now make their own database for their own barcoding region of choice, or download additional databases from the Metaxa2 Database Repository. The selection of other databases is made through the “-g” option already existing in Metaxa2. As part of these changes, we have also updated the classification scoring model for better stringency and sensitivity across multiple databases and different genes. The old scoring system can still be used by specifying the –scoring_model option to “old”.

There are two different main operating modes of the Metaxa2 Database Builder, as well as a hybrid mode combining the features of the two other modes. The divergent and conserved modes work in almost completely different ways and deal with two different types of barcoding regions. The divergent mode is designed to deal with barcoding regions that exhibit fairly large variation between taxa within the same taxonomic domain. Such regions include, e.g., the eukaryotic ITS region, or the trnL gene used for plant barcoding. In the other mode – the conserved mode – a highly conserved barcoding region is expected (at least within the different taxonomic domains). Genes that fall into this category would be, e.g., the 16S SSU rRNA, and the bacterial rpoB gene. This option would most likely also be suitable for barcoding within certain groups of e.g. plants, where similarity of the barcoding regions can be expected to be high. There is also a third mode – the hybrid mode – that incorporates features of both the other. The hybrid mode is more experimental in nature, but could be useful in situations where both the other modes perform poorer than desired.

In the divergent (default) mode, the database builder will start by clustering the input sequences at 20% identity using USEARCH (2). All clusters generated from this process are then individually aligned using MAFFT (3). Those alignments are split into two regions, which are used to build two hidden Markov models for each cluster of sequences. These models will be less precise, but more sensitive than those generated in the conserved mode. In the divergent mode, the database builder will attempt to extract full-length sequences from the input data, but this may be less successful than in the conserved mode.

In the conserved mode, on the other hand, the database builder will first extract the barcoding region from the input sequences using models built from a reference sequence provided (see above) and the Metaxa2 extractor (1). It will then align all the extracted sequences using MAFFT and determine the conservation of each position in the alignment. When the criteria for degree of conservation are met, all conserved regions are extracted individually and are then re-aligned separately using MAFFT. The re-aligned sequences are used to build hidden Markov models representing the conserved regions with HMMER (4). In this mode, the classification database will only consist of the extracted full-length sequences.

In the hybrid mode, finally, the database builder will cluster the input sequences at 20% identity using USEARCH, and then proceed with the conserved mode approach on each cluster separately .

The actual taxonomic classification in Metaxa2 is done using a sequence database. It was shown in the original Metaxa2 paper that replacing the built-in database with a generic non-processed one was detrimental to performance in terms of accuracy (1). In the database builder, we have tried to incorporate some of the aspects of the manual database curation we did for the built-in database that can be automated. By default, all these filtration steps are turned off, but enabling them might drastically increase the accuracy of classifications based on the database.

To assess the accuracy of the constructed database, the Metaxa2 Database Builder allows for testing the detection ability and classification accuracy of the constructed database. This is done by sub-dividing the database sequences into subsets and rebuilding the database using a smaller (by default 90%), randomly selected, set of the sequence data (5). The remaining sequences (10% by default) are then classified using Metaxa2 with the subset database. The number of detections, and the numbers of correctly or incorrectly classified entries are recorded and averaged over a number of iterations (10 by default). This allows for obtaining a picture of the lower end of the accuracy of the database. However, since the evaluation only uses a subset of all sequences included in the full database, the performance of the full database actually constructed is likely to be slightly better. The evaluation can be turned on using the “–evaluate T” option.

Metaxa2 2.2 also introduces the database repository, from which the user can download additional databases for Metaxa2. To download new databases from the repository, the metaxa2_install_database command is used. This is a simple piece of software but requires internet access to function.

References

1. Bengtsson-Palme J, Hartmann M, Eriksson KM, Pal C, Thorell K, Larsson DGJ, Nilsson RH: Metaxa2: Improved Identification and Taxonomic Classification of Small and Large Subunit rRNA in Metagenomic Data. Molecular Ecology Resources (2015). doi: 10.1111/1755-0998.12399 [Paper link]
2. Edgar RC: Search and clustering orders of magnitude faster than BLAST. Bioinformatics, 26, 2460–2461 (2010).
3. Katoh K, Standley DM: MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Molecular Biology and Evolution, 30, 772–780 (2013).
4. Eddy SR: Accelerated profile HMM searches. PLoS Computational Biology, 7, e1002195 (2011).
5. Richardson RT, Bengtsson-Palme J, Johnson RM: Evaluating and Optimizing the Performance of Software Commonly Used for the Taxonomic Classification of DNA Sequence Data. Molecular Ecology Resources, 17, 4, 760–769 (2017). doi: 10.1111/1755-0998.12628