Review for "Real-time strain typing and analysis of antibiotic resistance potential using Nanopore MinION sequencing"

Completed on 25 Jul 2015

If you would like to claim attribution for this anonymous review, please contact Academic Karma.

Login to endorse this review.


Comments to author

Author response is in blue.

This manuscript describes the application of the MinION sequencer for analyzing bacterial genomes, namely the genomes of three pneumonia strains. The authors conducted the MinION sequencing of the pneumonia genomes which were extracted from in vitro cultured strains. The authors conclude that 30 minutes sequencing time could give sufficient information to identify bacterial species and strains and further two and twelve hours sequencing could identify "initial" and "complete" drug-resistance profiles, respectively. Although I admit that the application of the MinION sequencing is in the right direction towards developing a rapid diagnosis of infectious diseases, thus it would give enormous benefits for selecting proper therapeutic means. Indeed, immediate responses are essential for many lethal pathogens, not only pneumonias but also other bacteria and viruses. However, here I have to suggest that much more extensive analyses should be done in order to

demonstrate that this approach is truly possible for practical diagnostic purposes. Trials are needed actually using the clinical samples. Also, the extensive analyses should cover further in-depth bioinformatics characterization of the data in terms of the sequencing fidelity, the data coverage and their robustness and so on against the genes of interest or the entire genomes.

Clinical samples are out of the scope of this paper, however we have included a new mixed samples of S. aureus (25%) and E. coli (75%). We agree that clinical samples are the next step.

Major points;

1. Although I admit that the time needed for the sequencing step itself is important, the time for the sample preparation should be even more crucial. When applied to the clinical samples, how the authors would prepare the sample genomes? If this step takes hours or even days, their claim on the short sequencing time would not make much difference. Also, there seem several alternative methods such as those utilizing PCR in order to prepare the template, at the same time enriching the information of the genes of particular importance. At least pros and cos of possible methods should be discussed.

This is a good point. The current paper focuses on cultured samples, and as a result, this is out of scope. We agree that working on clinical samples is the next step.

2. Bioinformatics characterization of the data is generally insufficient. At least, depth and coverage of the sequence and their fidelity at the base level should be provided for the genes of interest as well as the entire genomes. Here, the coverage and representation of the sequences are particularly important. Assuming this is the shotgun sequencing of the genomes, the sequences appearing in the first 30 minutes should be different. Do the authors claim that the sufficient information should be robustly obtained from any trials? Simulation analyses by splitting the data should be needed, at least, to rational such a claim. Further rationales of the Bayesian inference they used (p.10) should be also described in the text.

We have provided much greater detail and justification for the Bayesian inference approach.

3. Citation of the previous studies is not sufficient. Importantly, a similar paper has been recently published, reporting the MinION analysis on Salmonella, in Genome Biology (http://www.genomebiology.com/content/pdf/s13059-015-0677-2.pdf). The authors should discuss the common and unique feature of their own paper by comparing the results with each other.

This paper is highly relevant, but was not published (or otherwise available) when our original manuscript was submitted. We have now updated our current manuscript to discuss this paper at various points in the manuscript, including in the introduction

"Several systems incorporating real-time analysis of MinION data have been developed recently such as the cloud based platform Metrichor (Oxford Nanopore), work by Quick et al~\cite{QuickAC2015} and
MetaPORE ~\cite{GreningerNF2015}, focusing on placing the sample on a phylogenetic tree but without providing an estimate of the confidence in this assignment."


and also in the discussion:

"Researchers have also recently reported that MinION sequencing data could accurately identify bacterial outbreak strains within 50 minutes of sequencing~\cite{QuickAC2015} by placing reads onto a phylogenetic tree; and drug resistance profile of a \sa{} sample determined using a de-Bruijn graph approach from 8 hours of sequencing data~\cite{BradleyGW2015}."

Minor points;

4. Figures are too much focused on the time-dependency of the sequencing accumulations. This could be concisely represented in a single figure. Instead, much mode data should be presented on the sequencing fidelity, coverage, robustness of the detection and so on.

Figure 2 shows the the sequencing yield versus time. We have added a new Figure 3 which shows the quality metrics of the sequencing runs. The other figures include the quantity of reads, and all inferences are now stated as a function of number of reads obtained as well as time taken to obtain those reads.

5. I could not understand Table 3. What are the "sensitivity"/"specificity" and "TP/FP"?

If there are several classes of the genes related to a particular drug resistance, is not it nonsense to describe them collectively? Or did the authors want to describe only the "presence" or "absence" of the drug resistance genes, where the presence of any should lead to the drug resistance? I also wonder if there are any SNPs related to the drug resistance. If any, could they also be considered by this approach?

We have modified the way in which these results are reported so that genes are reported rather than classes, so that now the sensitivity and specificity (as well as TP and FP) refer to detection of the presence of antibiotic resistance genes which were detected in MiSeq sequencing of these samples. The presence of an antibiotic resistance gene does not necessarily confer drug resistance. In this paper we only attempted to look at the potential resistance via annotation of the presence of genes.

6. Figure 3: What does the error bar mean?

The error bar in the figure (now Figure 4) represents the 95% confidence intervals in the estimate of the proportion of each species present. This is now made clearer in the figure legend

7. It seems that format of the abstract does not fit that of {{JOURNAL}}.

We now conform to the requirements of {{JOURNAL}} (Methods).

REVIEWER'S RESPONSES TO QUESTIONS

Does the work include all necessary controls?

If not, please comment on the additional controls that are required.

No: See below (major point 2).

Are the conclusions drawn adequately supported by the data shown?

If not, please explain.

No: See below (major point 1).

Are sufficient details provided to allow replication and comparison with related analyses that may have been performed?

No: See below (major point 3).

Does the manuscript adhere to the field standards for experimentation, nomenclature and public availability of data (or any other significant standards)? Is the software freely available, open source and with an appropriate free-to-use license?

Yes

Does the method perform better than existing methods (as demonstrated by direct comparison with available methods)?

No: See below (major point 3).

Is the method likely to be of broad utility? Is any software component easy to install and use?

Please indicate briefly the novel features and/or advantages of the method, and/or please reference the relevant publications and which methods, if any, it should be compared with.

Potentially, this method should be of broad utility. However, it is not fully demonstrated here.

We now present extra data, and more detail on the methods we have developed.

Is the paper of broad interest to others in the field, or of outstanding interest to a broad audience of biologists?

If yes, please explain why.

Yes: Rapid diagnosis of pathogens and their drug sensitivity based on this method, if possible, would greatly contribute to the proper treatment of the patients.