Jean Cury and colleagues (2021) have developed a protocol to simulate bacterial evolution in SLiM. In contrast to existing methods that depend on the coalescent, SLiM simulates evolution forward in time. SLiM has, up to now, mostly been used to simulate the evolution of eukaryotes (Haller and Messer 2019), but has been adapted here to simulate evolution in bacteria. Forward-in-time simulations are usually computationally very costly. To circumvent this issue, bacterial population sizes are scaled down. One would now expect results to become inaccurate, however, Cury et al. show that scaled-down forwards simulations provide very accurate results (similar to those provided by coalescent simulators) that are consistent with theoretical expectations. Simulations were analyzed and compared to existing methods in simple and slightly more complex scenarios where recombination affects evolution. In all scenarios, simulation results from coalescent methods (fastSimBac (De Maio and Wilson 2017), ms (Hudson 2002)) and scaled-down forwards simulations were very similar, which is very good news indeed.
A biologist not aware of the complexities of forwards, backwards simulations and the coalescent, might now naïvely ask why another simulation method is needed if existing methods perform just as well. To address this question the manuscript closes with a very neat example of what exactly is possible with forwards simulations that cannot be achieved using existing methods. The situation modeled is the growth and evolution of a set of 50 bacteria that are randomly distributed on a petri dish. One side of the petri dish is covered in an antibiotic the other is antibiotic-free. Over time, the bacteria grow and acquire antibiotic resistance mutations until the entire artificial petri dish is covered with a bacterial lawn. This simulation demonstrates that it is possible to simulate extremely complex (e.g. real world) scenarios to, for example, assess whether certain phenomena are expected with our current understanding of bacterial evolution, or whether there are additional forces that need to be taken into account. Hence, forwards simulators could significantly help us to understand what current models can and cannot explain in evolutionary biology.
Cury J, Haller BC, Achaz G, Jay F (2021) Simulation of bacterial populations with SLiM. bioRxiv, 2020.09.28.316869, version 5 peer-reviewed and recommended by Peer community in Evolutionary Biology. https://doi.org/10.1101/2020.09.28.316869
De Maio N, Wilson DJ (2017) The Bacterial Sequential Markov Coalescent. Genetics, 206, 333–343. https://doi.org/10.1534/genetics.116.198796
Haller BC, Messer PW (2019) SLiM 3: Forward Genetic Simulations Beyond the Wright–Fisher Model. Molecular Biology and Evolution, 36, 632–637. https://doi.org/10.1093/molbev/msy228
Hudson RR (2002) Generating samples under a Wright–Fisher neutral model of genetic variation. Bioinformatics, 18, 337–338. https://doi.org/10.1093/bioinformatics/18.2.337
please revise your manuscript according to the reviewers' comments. Please make sure to address and reply to every comment.
Thank you for submitting the revised version of the manuscript. Although, I feel like the current version is an improvement, I cannot recommend it for the following reasons:
1. In your reply you wrote “The adaptation of SLiM presented here is not aimed at competing with FastSimBac or ms on “simple” scenarios but rather to open new simulation possibilities”. Indeed this was not clear in the last version you submitted. One of the main reasons this was not clear is that you failed to present data to support the point that SLIM opens up new simulation possibilities. It is indeed important to compare the SLIM code to existing methods but what needs to follow is a detailed analysis of a novel use case or a novel class of use cases that are impossible or difficult to simulate with existing methods or a use case that simply solves an interesting biological problem. The use case you present now, seems to be an interesting one. Yet, currently there is only a brief mention and a figure of the results. There is no analysis and no code to replicate the figure. The reviewers and I feel like the value of the paper really hangs on the detailed analysis of such an example. Without it, the additional value the manuscript provides over existing SLIM manuscripts and existing simulation methods is small. In my opinion this is also the best way to achieve the aim you state at the end of the discussion “We hope that our work here will stimulate a wave of development of simulation-based models for bacterial population genetics.”. Scientists will certainly be animated by an amazing analysis of a simulated evolution experiment that solves an actual biological question. Like it has been done with other scripting languages such as Avida.
2. Unfortunately you have not replied to many points the reviewers have made. For example, one of my comments has been left unanswered. Why are there negative values when you normalize the results? I can see that they have now disappeared, but what happened? Also, Supplementary Figure 11 still has those negative values. Is this intended?
3. Supplementary Figures are not numbered correctly and some Supplementary Figures in the text do not seem to exist (or maybe they exist but have a different number).
4. Supplementary Figure 8 is impossible to understand. Why does A start with 2? What does burn-in through capitation mean? What is shown in the figure (e.g. what are the different colored circles?)?
5. Generally, the figure legends have not improved unfortunately. I wish I could be more positive, but in its current state I cannot recommend the submitted manuscript.
The reviewers find the approach presented here interesting, but criticize that you have not established the specific advantage of the presented approaches over existing approaches. We feel it is important to analyse common use cases where existing approaches fail or cannot be applied. So far the only comment on the advantage of SLiM over the other methods seems to be that SLiM can take circular genomes into account (How much does this matter?). Furthermore, we find it difficult to interpret the data and figures presented in the results section. For example, the data presented in Figure 2: 1. There is no 1 to 1 comparison between the WF expectation and the simulation results. For example, a simulation without recombination would be useful to show that in ideal circumstances the simulations perform as expected. 2. Why/how can the normalization lead to negative values? A better explanation of how the normalization works would be helpful interpreting the figure. It is also unclear what exactly the figures are intended to show. If the main aim of the figure is to show that rescaling does not have an effect on the data, then the figure should show a direct comparison between different scaling factors. Once it is established that the scaling factors do not change the results, SLiM could then be compared to existing methods. In general, as has been pointed out by the reviewers, improved figure legends would help with understanding the presented data. Finally, jargon and abbreviations are used to an extent that the paper becomes difficult to read. In conclusion the manuscript requires very substantial revision in order to be recommended. Importantly, we feel a revision should include data regarding the advantage of SLiM over existing methods.
Additional requirements of the managing board:
As indicated in the 'How does it work?’ section and in the code of conduct, please make sure that:
-Data are available to readers, either in the text or through an open data repository such as Zenodo (free), Dryad or some other institutional repository. Data must be reusable, thus metadata or accompanying text must carefully describe the data.
-Details on quantitative analyses (e.g., data treatment and statistical scripts in R, bioinformatic pipeline scripts, etc.) and details concerning simulations (scripts, codes) are available to readers in the text, as appendices, or through an open data repository, such as Zenodo, Dryad or some other institutional repository. The scripts or codes must be carefully described so that they can be reused.
-Details on experimental procedures are available to readers in the text or as appendices.
-Authors have no financial conflict of interest relating to the article. The article must contain a "Conflict of interest disclosure" paragraph before the reference section containing this sentence: "The authors of this preprint declare that they have no financial conflict of interest with the content of this article." If appropriate, this disclosure may be completed by a sentence indicating that some of the authors are PCI recommenders: “XXX is one of the PCI XXX recommenders.”