Challenges Using Droplet Digital PCR for Environmental Samples

Round 1

Reviewer 1 Report

This is a MS submitted to Applied Microbiology exploring methodological recommendations regarding ddPCR. Although I found the information is useful, I think the MS needs some improvement so broad readers especially those that are not familiar with ddPCR can grasp the key messages. 1. In the beginning the authors indicated that they included multiple biological replicates (line 101-109). Is all of the graphs consist of average/pooled analysis from multiple biological replicates? For instance, in Fig 2, the sample codes were written A02-F02 and then the graph after the X-axis label changes to B01-F02 2. My understanding in relation to microbial quantification using PCR is, a user can quantify gene coy number of targeted taxa. In this MS, how the author confirms that their optimization could increase the efficiency of detection to reach the “true” number of the biological samples? 3. In my opinion, this MS is lacking a statistical analysis component. I don’t get an impression that sufficient statistical analysis was performed. Is the result statistically different between the optimization approach or running condition? 4. For a reader that is not familiar with the ddPCR, the idea of introducing this technique is interesting. However, I can not grasp the key findings on how this system is compared to qPCR which is a more known and commonly used technique? Why the author did not perform a comparison regarding the efficiency between these two techniques? Minor 1. Some grammatical error i.e. line 198, line 269 2. What is the target (gene) of primer used in line 112-119. 3. I am just wondering in relation to “optimizing” cycling condition – what is the justification to increase the cycle to 50. My understanding in relation to normal PCR, more than 45 cycles is not recommended as nonspecific amplification start to appear

Author Response

Reviewer 1

Thank you very much for you feedback in our work.

 

This is a MS submitted to Applied Microbiology exploring methodological recommendations regarding ddPCR. Although I found the information is useful, I think the MS needs some improvement so broad readers especially those that are not familiar with ddPCR can grasp the key messages.

 

1. In the beginning the authors indicated that they included multiple biological replicates (line 101-109). Is all of the graphs consist of average/pooled analysis from multiple biological replicates? For instance, in Fig 2, the sample codes were written A02-F02 and then the graph after the X-axis label changes to B01-F02

** We deliberately chose environmental samples that had a lot of rain or were problematic when using the conventional ddPCR protocol. We clearly state this now at the beginning of our results section L245 which is also underlined. Please keep in mind that the sample codes A02 etc. are just the plate wells that the samples were placed in and do not represent the sample perse. It is just the location to the ddPCR plate (we also explain that in our Supplemental Figure 1). To put it simply, the same sample can have a different well name and depends on which well it was placed during the analysis.

 

2. My understanding in relation to microbial quantification using PCR is, a user can quantify gene copy number of targeted taxa. In this MS, how the author confirms that their optimization could increase the efficiency of detection to reach the “true” number of the biological samples?

** The biggest problem in ddPCR is how to produce “clean” environmental samples and set a proper threshold in order to accurately quantify the target. This entails a few specific steps. 1) You need to limit the “rain” (as we explain in L52 – 64). Samples with lot of rain do not allow for proper target quantification. 2) You need to ensure that the threshold properly separates the negative cloud (not amplified droplets and non-specific amplification droplets) from the positive cloud (amplified droplets with your target) (as we describe in L76-83).

• Our work provides the necessary cycling optimization steps that need to be applied for environmental samples in order to limit rain. And in the case of persistent rain, we provide additional advice on how to discriminate between true and false positives (e.g., spiking with positive).
• Also, our work describes the appropriate controls that need to be used in environmental studies in order to properly set the threshold and accurately optimize the target.

 

The reduction of rain and the exclusion of false positives using proper controls and PCR optimization demonstrates the efficiency of the protocol. As an example, you can also see figure 7 as an example which shows how improper threshold skews the results.

 

Overall, the reduction of rain and the discrimination between false and true positives is a result by itself increasing the efficiency of “true” target quantification in unknown environmental samples.

 

3. In my opinion, this MS is lacking a statistical analysis component. I don’t get an impression that sufficient statistical analysis was performed. Is the result statistically different between the optimization approach or running condition?

**Please keep in mind that this is a protocol optimization study that aims to provide the necessary steps for 1) rain limitation and 2) proper threshold determination as we mentioned in the previous response. Reduction in rain after using our protocol (as we mention in L251 -252) from 2394 to 254 droplets demonstrates the importance of our work. Also, identification of false positives via the use of proper controls is also an absolute result (see figure 6). Adding a p-value to these results is not needed (which would be highly significant regardless).

 

4. For a reader that is not familiar with the ddPCR, the idea of introducing this technique is interesting. However, I can not grasp the key findings on how this system is compared to qPCR which is a more known and commonly used technique? Why did the author not perform a comparison regarding the efficiency between these two techniques?

**We understand your concern and that is why we include a brief comparison between the two systems in our introduction. But the scope of our work is not to prove ddPCR better (this topic has been covered extensively byt other authors and by BioRad itself (that provides both instruments). In comparison to qPCR, ddPCR does not require sample replication (qPCR needs 3 reps per sample), has a better target detection limit (can pick up rare targets) and does not need to generate a standard curve in order to quantify your target rather it gives absolute quantification per sample right away. To help readers we now include that information in the figure 1 caption where the 2 systems are compared. Also, as we mention in the caption, optimization for qPCR for environmental samples has been already published (please see ref #10) but is missing for ddPCR. Finally, since both qPCR and ddPCR are extremely expensive instruments usually only one of the two is available in labs/institutions. 

 

Few example studies that tackled this question:

“Comparison of next-generation droplet digital PCR (ddPCR) with quantitative PCR (qPCR) for enumeration of Cryptosporidium oocysts in faecal samples”

“Droplet Digital PCR versus qPCR for gene expression analysis with low abundant targets: from variable nonsense to publication quality data”

“A multiplexed droplet digital PCR assay performs better than qPCR on inhibition prone samples”

 

Minor 1. Some grammatical error i.e., line 198, line 269

** We corrected the errors, please see all highlighted sections in our revised manuscript

 

2. What is the target (gene) of primer used in line 112-119.

** This is a very good point, we now included information on the targeted gene: Please see L121 – L123 “The primers and probes used are specific to R. irregulare strain DAOM197198 targeting the cox3-rnl intergenic mtDNA region and the sequences are:”

 

3. I am just wondering in relation to “optimizing” cycling condition – what is the justification to increase the cycle to 50. My understanding in relation to normal PCR, more than 45 cycles is not recommended as nonspecific amplification start to appear

** you are absolutely right, that was a mistake in the table. We examined the increase of PCR cycles from 40 to 45 but not beyond as we mentioned in the material and methods section L148 – 149. The mistake on the table has now been corrected, thank you for noticing this.

 

 

 

 

Reviewer 2 Report

Dear authors, please find my comments and suggestions attached.

Thank you for your work

Comments for author File: Comments.pdf

Author Response

Reviewer 2

Thank you very much for your suggestions and for your time.

 

**We corrected all the typos pointed out in the manuscript. Please see the highlighted areas in our revised manuscript.

 

L35: what about environmental samples? It is worth mentioning earlier in the manuscript that despite its limitations, ddPCR still represents a better option than classical qPCR for environmental samples as well

** Thank you for your suggestion, we included this in our abstract (and in figure 1 caption), so the reader are aware early on that ddPCR is appropriate and better than qPCR for environmental samples. Please see L13-14

 

Explain why spores would be comparable to soil inhibitors

** In fact, they act as the inhibitor free control. We included that explanation now in the appropriate section. Please see L114-L117 “The organismal positive control is used to demonstrate the optimal amplification of our target when not affected by inhibitors that are usually present in environmental samples and also serves as an indicator for the primers/probes functionality.”

 

This figure is very useful to the reader and should be moved earlier in the manuscript, for results interpretation. Consider making it an actual figure instead of supplementary

** Because of the size of the figure and the figure caption we still think that it is better to be in the SM (unless the editorial team also agrees that it can be in the main manuscript). But because you pointed out it is important to look at before the result section, we moved the reference to Figure S1 to the beginning of our result section.

 

**Figure 6: We included information for all the samples presented in the figure caption.

Round 2

Reviewer 1 Report

My comments from the previous revision are adequately addressed. I recommend to accept in present form.

Author Response

Thank you very much for you comments that imporved our manuscript.