Check Claims Before You Publish

 Quick Explanation Copied

 Long Explanation

Citation Metrics: Two Irreconcilable Records

The supplied metadata for "Tom Blundell" is internally contradictory and must be reconciled before scoring. The Semantic Scholar-style record lists an h-index of 3, 112 citations, and 6 papers — including titles such as "Plane common sense" and "Energy: the changing climate" (a Royal Commission report) — which are clearly unrelated to structural biology and almost certainly belong to a different Tom Blundell. The OpenAlex record for Tom L. Blundell (ORCID 0000-0002-2708-8992) instead reports 730 works, 63,029 citations, and an h-index of 114, with topic scores dominated by Biology (0.96), Chemistry (0.93), and Computational Biology (0.75). The evidence-weighted conclusion is that the OpenAlex record is the correct identity for the Cambridge structural biologist, and the low-citation record is an entity-resolution artifact — a caution against taking bibliometric databases at face value.

The 1993 spike reflects the foundational comparative-modelling paper with Šali (top work, 13,000+ citations), and the 2015 spike the pkCSM pharmacokinetics predictor (5,717 citations) — both field-defining methods papers.

Recent Primary Work: HANSEN (2026)

The most recent supplied paper, the HANSEN resource for Mycobacterium leprae, integrates AlphaFold 3, Boltz-1/2, and Chai-1 models across 1,603 proteins, with monomer models benchmarked against all seven available experimental structures (mean TM-scores 0.82–0.90; median Cα-RMSD 1.9 Å) HANSEN M. leprae structural proteome resource. The authors explicitly frame outputs as hypothesis-generating predictions requiring experimental validation, and disclose that essentiality relies on cross-species transfer from M. tuberculosis because M. leprae is experimentally intractable — a mark of epistemic honesty, though only seven reference structures limits the accuracy benchmark.

Breadth and Rigor Across the Record

The supplied portfolio spans computational structural biology (environment-specific substitution tables improving active-site prediction across 339 SCOP domains ESST functional-residue masking study), careful biophysics (pH-tuneable ligand binding in ketopantoate reductase, crystallography at 1.95 Å plus ITC with mutants KPR pH-dependent ligand binding study), and even condensed-matter magnetism Copper-pyrazine antiferromagnet study — though the Physics paper likely reflects a different co-author of the same name, another disambiguation caution. The self-assessment limitation pattern in the HANSEN paper (pLDDT as self-assessment, ranking as triage not validation) demonstrates consistently calibrated interpretation rather than over-claiming.

Confidence note: High confidence on the OpenAlex-based career metrics; the Semantic Scholar record is judged unreliable due to evident identity confusion, and this remains the largest blind spot in the supplied data.

 BGPT Author Review

Scientific Quality

OpenAlex metrics (h-index 114, ~63k citations, 730 works) plus authorship of two of the most-cited methods in computational structural biology (MODELLER, pkCSM) place him at the top of the field. Recent work shows modern method integration with honest limitation disclosure. The conflicting low-metric record is almost certainly a disambiguation error, not evidence of weak science.

Communication Quality

Supplied papers pair quantitative benchmarking with clearly stated caveats and accessible framing (e.g., predictions 'for hypothesis generation'). Weakness: recent resource papers are dense multi-method descriptions where interpretive synthesis is delegated largely to supplementary tables.

Author Novelty

Created foundational protein-modelling and mutation-effect prediction tools that spawned entire subfields; recent HANSEN work applies multi-model AI prediction to an underserved pathogen proteome, which is incremental in method but novel in application scope.

Scientific Rigor

Consistent use of external benchmarks (TM-scores against experimental structures, homology-grouped cross-validation, ITC with point mutants), explicit falsification-relevant framing, and self-declared limitations such as pLDDT being self-assessment. Minor weakness: HANSEN's essentiality validation rests on only 39 labelled transfer proteins and cross-species orthology.