Escribir un comentario

Major Issues

1. The sampling strategy is under-described and likely biased. The study uses only 89 Delta genomes from India and 74 Omicron genomes from South Africa, selected from early GISAID submissions. The manuscript does not explain how these sequences were chosen, whether duplicate/outbreak-linked sequences were removed, whether sampling dates and locations were balanced, or how representative they are of national variant spread. Because phylodynamic estimates are very sensitive to sampling, this is a major limitation.

2. Delta and Omicron are compared across different countries and time periods. The analysis compares Delta in India with Omicron in South Africa. Differences in estimated Re, doubling time, and effective population size may reflect country-specific surveillance intensity, interventions, immunity, testing, demography, and sampling effort rather than intrinsic variant biology. The authors should frame the results as a comparison of two early country-specific outbreaks, not as a direct general comparison of Delta versus Omicron.

3. Some model descriptions and parameter choices are unclear or inconsistent. The manuscript reports using both BEAST v1.10.4 and BEAST v1.8.4, then BEAST2 v2.6.3 for birth-death skyline. It also states HKY+4, likely meaning HKY+G or HKY+I/G, and reports a fixed evolutionary rate of 0.001 subs/site/year for birth-death analysis despite estimating different rates for Delta and Omicron. These details need clarification because they affect reproducibility and interpretation.

4. The birth-death skyline priors require stronger justification. The chosen priors for Re, sampling probability, origin time, and becoming-uninfectious rate are not sufficiently justified using epidemiological evidence. The infectious-period priors and resulting rates should be checked carefully, because the reported units and conversions are confusing. Sensitivity analyses using alternative priors would strengthen confidence in the estimates.

5. Uncertainty is often too large for the strength of the conclusions. Several estimates have very wide HPD intervals. For example, Delta tMRCA spans from late October 2020 to late March 2021, and Re intervals include values below and above 1. Omicron Re also has a broad interval. The conclusions should be softened to reflect uncertainty rather than stating that dates “coincide perfectly” or that Omicron spread “three times faster” as a firm result.

6. Figures appear missing from the PDF. The PDF includes figure captions for Figures 1-3, but the rendered pages I checked show captions without the actual figure panels. If this is also true in the submitted version, the figures need to be restored. Without the skyline plots, readers cannot visually assess the evidence behind the key results.

7. The manuscript needs updated and more careful biological interpretation. The discussion attributes faster Omicron growth largely to genomic features, but the analysis does not directly test immune escape, prior immunity, vaccination, behavior, interventions, or ascertainment. The authors should distinguish between phylodynamic estimates and mechanistic explanations.

Minor Issues

1. The title should be revised for grammar: “Comparison of the evolutionary phylodynamics of Delta and Omicron variants of SARS-CoV-2” would read better.

2. The abstract repeats the doubling-time conclusion twice and should be tightened.

3. “Birth-death horizon” should likely be “birth-death skyline.”

4. Variant names should be styled consistently: Delta and Omicron, rather than DELTA and OMICRON throughout.

5. The manuscript contains many language and formatting issues, including missing characters, spacing errors, and unclear symbols. A careful copyedit is needed.

6. The formula for doubling time appears missing after “estimated by the equation.” Please include the actual equation.

7. The table should define all units clearly, especially for lambda, delta, and sampling probability.

8. The “sampled population” estimates derived from rho are confusing and need clearer explanation.

9. The methods should provide GISAID accession IDs or an acknowledgment table so the dataset can be reproduced.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.