Check Claims Before You Publish

Visual summary (graphs first)

Evidence synopsis (claims & sources)

1. Research focus and methods: Seuma’s work centers on multiplexed assays of variant effects (MAVEs / DIM) applied to amyloid‑β and prion‑like domains, combining large synthetic variant libraries, yeast selection assays, deep sequencing, and computational analyses to map sequence→nucleation relationships at scale; raw data and pipelines are openly provided in GEO and GitHub for reproducibility (see Nature Communications DIM atlas; eLife Aβ DMS) Nature Comm DIM‑Aβ Atlas
2. Biological conclusions: Multiple independent studies she co‑authored show that (a) many indels and truncations—particularly in the N‑terminal region of Aβ42—accelerate nucleation, (b) deep mutational maps discriminate known familial Alzheimer’s mutations and identify hotspots (e.g., E22Δ, Δ19‑24), and (c) minimal truncated cores combined with N‑terminal positive charge can strongly increase nucleation kinetics; these conclusions are supported by high sequencing depth and replication but limited by the yeast surrogate assay and peptide length scope eLife Aβ DMS mapping fAD
3. Reproducibility & openness: High: data (GEO), processed tables, and analysis code are available (DIM atlas repository); sequencing depth and replicate structure indicate robustness to sampling noise, and multiple downstream analyses (ROC/AUC, predictor benchmarking) are provided in the manuscripts and repositories Nature Comm DIM‑Aβ Atlas (reproducibility)

Critical appraisal — strengths, limitations, and blindspots

• Strengths: rigorous experimental design (large library sizes, deep sequencing, multi‑replicate selection), open data/pipelines, clear mechanistic insights relating sequence regions to nucleation, and articles published in high‑quality journals (Nature Communications, eLife) supporting external validation and community uptake.
• Limitations / blindspots: primary assays use a yeast Sup35N‑Aβ42 fusion surrogate and a short peptide (Aβ42), which may not capture neuronal cellular context, post‑translational processing, or longer isoform behavior; NS scores report selection enrichment rather than absolute in‑vitro kinetic constants, and some generalization to other amyloids remains to be shown explicitly DIM atlas limitations
• Potential biases to watch: yeast surrogate selection may introduce context‑dependent biases (differences in chaperones, proteostasis), publication bias toward positive/interesting variants, and interpretation gaps between NS enrichment and human pathogenicity; the authors openly discuss many of these caveats in their papers (good practice).

Practical interpretation for peers and reviewers

If you evaluate Seuma’s work for collaboration, grant review, or reproducibility: the datasets are high quality (deep coverage, replication), methods and code are available, and the biological conclusions are well supported within the experimental system; however, any translational claim toward human neuronal pathology should be treated conservatively unless corroborated by orthogonal biochemical kinetics or mammalian cellular models.

Key documents & direct evidence (open links)

• DIM Aβ42 Atlas (full text PDF)
• eLife DMS Aβ substitutions
• Prion-like domain MAVE (Nature Communications, 2019)

What would change this assessment (falsification criteria)

• Independent replication of the DIM atlas in an orthogonal mammalian neuronal model showing no enrichment of NS+ variants in the N‑terminus would undermine system‑specific generalization.
• Direct biochemical measurements (primary/secondary nucleation kinetics) across a broad set of DIM‑identified NS+ variants showing no correlation with NS scores would weaken the functional translation of selection enrichment to nucleation kinetics.

Bottom line (evidence‑weighted)

Mireia Seuma is a productive, methodologically rigorous researcher in the MAVE/DIM space with strong openness and reproducibility practices and multiple contributions that materially advance understanding of sequence determinants of amyloid nucleation; her results are robust within high‑quality yeast selection frameworks but require orthogonal validation before making broad translational claims about human pathophysiology.