Comments on abstract, title, references

The title of the paper suits the description of the paper

The paper describes a toolset to predict the effects of variants. It provides different options and configurations to extend the variant-based analyses, hence will be very helpful to accelerate the variant interpretation process.

References are relevant and cite appropriate studies and resources to support the claims and also enrich the knowledge of the audience

Comments on introduction/background

The authors have clearly explained the challenges posed by ever-increasing sequencing data, evolving annotation schemes, which affect the interpretability of variants. A comprehensive account of the comparison of the newly proposed toolset with the two popular tools, Annovar and SnpEff, points out the lacunae in existing ones and the scope for the improvements in the variant annotations.

Comments on methodology

The algorithm for variant annotation and effect prediction is well described. The software is made freely available in the form of an online web server and standalone scripts hence can easily be accessed by anyone. This also enables the readers to assess the reproducibility of the prediction results. The algorithm is also provisioned with QC checks and log collection methods to further investigate the failed/missing annotations for a subset of variants.

Comments on data and results

The results provide various levels of details w.r.t. input and output by VEP. The output fields are clearly explained with interpretation guidelines and appropriate references. VEP output includes variant annotations at transcript, protein, non-coding levels and also provides information from other resources such as 1000 genome project, ClinVar, ExAc, HGMD, dbSNP, etc. This is truly a set of enriched annotations for input variants and enables variant interpretation in sufficient depth.

Authors have ensured the ease of accessing the VEP by implementing it using both web server and standalone Perl package, and also made it available in the form of REST APIs. The suite of VEP plugins further enhances the accessory usage of tools.

The source code VEP is freely available on GitHub with a CI/CD framework incorporating a range of test cases that assure the functionality of the package during the course of active development.

Comments on discussion and conclusions

The runtime performance of the VEP is compared with SnpEff and Annovar. Given the depth of variant annotation, the runtime of VEP-based analysis is justified. Comparison of computational memory requirements would have improved the comparative account.

VEP is open to new developments in the variant resources and standards, such as GTEX, and GA4GH. The authors have also pointed out the limitation of the current variant annotation methods as they do not consider the combined effect of variants at multiple loci on the phenotype/trait.