Review papers with raw data transparency

Visual summary — measured single‑site coverage and focus of DMS datasets discussed in the paper

Visualized takeaways (figures → evidence)

• Primary synthesis: The review synthesises high‑coverage DMS maps that identified RBD positions strongly enhancing ACE2 binding (N501Y, Q498R/ H context‑dependence) and immune‑dominant escape regions around E484 — backed by the foundational RBD yeast display DMS datasets Starr et al., RBD DMS
• Protease landscapes and drug‑escape: Mpro DMS (>99% single‑site) and follow‑up resistance mapping predicted contact‑site and distal allosteric escape mutations and showed fitness tradeoffs; independent Mpro DMS work supports these conclusions Flynn et al., Mpro DMS
• PLpro mapping: The review highlights a near‑complete PLpro single‑site map in mammalian cells that combined activity and abundance assays to separate folding defects from catalytic loss, and identified S4 pocket flexibility and residues (e.g., M208) modulating inhibitor susceptibility Wu et al., PLpro DMS
• Diagnostics risk (N protein): DMS mapped nucleocapsid substitutions that abrogate binding by commercial rapid antigen test mAbs, implying overlapping epitopes across kits and potential diagnostic escape Frank et al., N DMS

Critical strengths

1. Concise, up‑to‑date synthesis tying multiple orthogonal DMS modalities to translational questions (surveillance, therapeutics, diagnostics) and pointing to concrete residue‑level predictions (supported by primary DMS studies cited above). Call et al., review summary
2. Balanced discussion about assay limitations (yeast vs mammalian glycosylation, permeability/drug‑efflux issues in yeast Mpro screens, epistasis from multi‑mutation backgrounds) and actionable recommendations (use variant‑background scans, replicates across assay systems).

Key limitations / blindspots the review correctly highlights (and where it could go further)

• Single‑site DMS cannot fully capture epistasis: the review notes this and references work showing shifting mutational constraints when RBDs are scanned on variant backgrounds — follow‑up combinatorial or background‑matched DMS is required to predict multi‑mutation trajectories reliably Starr et al., background dependence
• Assay context dependence: yeast glycosylation and mammalian overexpression can alter surface display or folding; the review notes but could have expanded with concrete benchmarking guidance (e.g., paired yeast/mammalian cross‑validation, direct viral replicon or infectious virus validation where safe/possible) — replicon approaches are explicitly recommended as future opportunities He et al., replicon
• Limited direct in vivo validation across many DMS hits: the minireview flags this translational gap but could have prioritized which DMS predictions already have orthogonal in‑culture/in vivo validation and which remain speculative.

Concrete, actionable recommendations (from re‑analysis of review + primary data)

1. For surveillance: integrate DMS positional escape maps into variant‑scoring pipelines (phylogenetic + epitope impact), but weight predictions by background sequence (epistasis) and assay system (yeast vs mammalian) Dadonaite et al., DMS + forecasting
2. For antiviral development: use DMS maps to prioritise inhibitor contact points that are functionally constrained (low tolerance) and to prospectively screen candidate inhibitors for escape via DMS‑based drug pressure assays, but validate resistance trajectories in viral passaging and clinical isolates (Mpro DMS predicted E166V/L50F pairings and subsequent compensatory evolution observed in passaged virus) Zhou et al., Mpro escape in culture
3. For diagnostics: DMS mapping of N protein epitopes should be integrated into regulatory test validation and surveillance to detect antigenic escape that could yield false negatives, prioritising multi‑epitope capture antibodies to reduce single‑mutation escape risk Frank et al., diagnostic escape

Where the review could be strengthened (brief)

• Stronger tabulated comparison of assay platforms (yeast vs mammalian vs pseudovirus vs replicon) with quantified biases and example concordance/discordance rates across matched variants.
• A short decision tree for when to trust a DMS‑derived prediction (e.g., concurrence across assay platforms, effect size, conservation, and phylogenetic recurrence in surveillance data).
• More explicit links between DMS scores and phylogenetic phylodynamic models (e.g., MFBD/Lumiere approaches) to quantify population‑level impact of measured fitness effects Rasmussen et al., MFBD

Confidence, falsifiability and what would change the conclusions

The review’s central claims (that DMS provided residue‑level predictions for ACE2 affinity, antibody escape, protease drug‑escape and diagnostic vulnerability) are well supported by primary DMS publications cited above. Conclusions would be overturned or substantially revised if:

• Large‑scale combinatorial DMS across circulating variant backgrounds showed consistent and frequent sign‑reversals of single‑site predictions (strong epistasis invalidating single‑site extrapolations).
• Multiple DMS hits (predicted to confer drug escape or antigenic escape) systematically failed to alter viral fitness or immune recognition in validated viral replicon / infectious virus and clinical isolate assays.

Interactive next steps

To iterate deeper on this review I can (pick any):

• Produce a tabular concordance matrix comparing RBD/Mpro/PLpro/N DMS effect sizes across assay platforms (requires raw DMS score files).
• Re-score surveillance sequences with DMS‑derived antigenic/drug escape weights and produce a ranked list of current variants of concern for surveillance prioritisation.