First edits

[tram-nguyen-n]

Hi @lgeistlinger would you mind giving the package a first-look and provide me with any edits you may have please?

[lgeistlinger]

Thanks @tram-nguyen-n - a couple of notes on the README:

• ProteinGym link is dead - change to https://proteingym.org (note the missing https://)
• evaluating the performance of multiple variant prediction tools -> evaluating the performance of models predicting the effect of point mutations

[lgeistlinger]

Some more:

• resources from ProteinGym, generated by Notin et al. 2023 -> ProteinGym (Notin et al., 2023)
• Currently, this package provides -> The package provides
• (DMS) scores generated from single substitution mutations across 216 assays -> (DMS) scores from 216 assays measuring the impact of all possible amino acid substitutions across xx proteins, and
• predicted AlphaMissense pathogenicity scores for ~1.6M substitutions matching those in the ProteinGym DMS data -> AlphaMissense pathogenicity scores for the ~1.6 M substitutions in the ProteinGym DMS data

[lgeistlinger]

A couple of notes on the DESCRIPTION file:

[lgeistlinger]

• Title: Programmatic access to ProteinGym datasets in R/Bioconductor
• Version: change to 0.99.0 to comply with Bioc package versioning scheme
• Description: match with description in the README
• some additional BiocViews to consider: CellCulture, SequencingData

[lgeistlinger]

Add installation instructions to README (for the moment only describe how to install the package from github) ...

[lgeistlinger]

There is currently no vignette. Bioconductor packages are required to have a vignette.

[lgeistlinger]

Functions names are supposed to start lower case according to Bioconductor coding style. Consider renaming:

• AlphaMissense_scores -> am_scores
• DMS_substitutions -> dms_scores

[lgeistlinger]

Function name mismatch in DMS_substitutions.R:

• Function name in function definition: DMS_substitutions
• Function name in example: ProteinGym_DMS_subs

[lgeistlinger]

DMS_substitutions currently returns a list of tibbles. AlphaMissense_scores returns an ordinary data.frame. Is there a strong argument for returning tibble and thus requiring the extra dependency? My preference would be for DMS_substitutions to return a list of ordinary data.frames.

[lgeistlinger]

' Load AlphaMissense scores for ProteinGym variants -> #' AlphaMissense scores for ProteinGym variants

' Load ProteinGym DMS Substitution Scores -> #' ProteinGym Deep Mutational Scanning (DMS) Scores

[lgeistlinger]

You don't need to link the bulky reference in details here. Just say: "See references."

Same here.

[tram-nguyen-n]

Thanks for the edits @lgeistlinger! Incorporated them all.

For this comment:

Functions names are supposed to start lower case according to Bioconductor coding style. Consider renaming:

• AlphaMissense_scores -> am_scores
• DMS_substitutions -> dms_scores

I kept it dms_substitutions to distinguish it from the DMS indel scores that are provided in ProteinGym too.

I can overwrite the list object on Azure to be data.frames instead of tibbles after Aaron and Pascal's edits.

[tram-nguyen-n]

changes incorporated