Check Claims Before You Publish

 Quick Explanation Copied

 Long Explanation

Identity Disambiguation Is the Dominant Uncertainty

Cheng Peng is one of the most collision-prone names in bibliographic databases. The supplied Semantic Scholar profile shows an h-index of 2, 14 citations, and 24 papers spanning anesthesiology, orthopedics, MRI diagnostics, botanical polysaccharides, economics and civil engineering — a pattern characteristic of a merged, disambiguation-corrupted profile rather than a single scientist. By contrast, OpenAlex lists multiple distinct candidates, including one Cheng Peng (ORCID 0000-0003-3303-906X) with 1,080 works, 32,631 citations and h-index 87, and others exceeding h-index 100. These metrics are mutually incompatible, so no single citation profile can honestly describe the queried author without ORCID-level verification.

What the Claim-Level Records Actually Show

The evidence records describe a highly coherent research program centered on in-situ structural virology and AI methods for biology. The VACV portal study combines cryo-ET (7,979 particles, 46 cores), AlphaFold3-assisted model building, and 4KA/4KR lysine mutagenesis with three biological replicates to show a lysine-lined constriction compatible with ssRNA export VACV portal complex architecture. Companion cryo-ET work resolves on-virion influenza HA oligomers (dimers/pentamers/hexamers; interface-2 disruption abolishes virus rescue) On-virion HA oligomerization cryo-ET and temperature/receptor-driven spike activation in HCoV-229E (S2-loose rising from 17.4% to 28.0% at 33°C; 53.2% Gemini S-hAPN complexes) Coronavirus spike on-virion dynamics. Parallel AI-methods papers include Counterpart Calibration Theory for de novo peptide sequencing FDR control (empirical FDR 2.6–5.3% vs 5% target across nine species) CCT FDR control for de novo sequencing, the GDCL-NcDA framework (5CV AUC 0.94–0.98) GDCL-NcDA ncRNA-disease prediction, and LLM/agent benchmarks (CAMA, SoMe with 9.2M posts) CAMA causal math reasoning, plus computational materials science on bilayer CrBr3 Bilayer CrBr3 magnetism DFT.

Rigor, Blindspots and Confidence

Rigor is consistently strong: multi-model validation, public data/code deposition (EMDB/PDB accessions, GitHub, PRIDE), and mutation-rescue controls are hallmarks of the structural papers. Yet important limitations recur: reliance on AlphaFold3 for component assignment, in vitro core preparations that may not recapitulate in vivo dynamics, host-dependent glycosylation on egg-derived virions, and database-label dependence in FDR calibration Portal study limitations. The claimed corpus is also dominated by very recent preprints (2025–2026) with zero recorded incoming citations, so external validation is still pending. What would change this assessment: ORCID confirmation linking (or unlinking) these papers to the queried name; independent replication of the HA interface-2 and portal-pore mutagenesis results; or incoming citations either validating or contradicting the CCT calibration in practice. Verify Author Identity on BGPT Check Funding Conflicts Run AI Scientist Analysis Evolve Author Review Make Movie Make Poster

Feedback:   

Updated: September 21, 2026

 BGPT Author Review

Scientific Quality

Score held to 6 despite excellent paper-level quality (8–9 on records) because the queried profile is demonstrably confounded: the h-index-2/14-citation record aggregates unrelated fields, while a separate OpenAlex Cheng Peng shows h-index 87 with 32,631 citations. Without ORCID verification the genuine author's scientific strength cannot be confirmed; the attributed structural-virology/AI corpus is, on its own merits, world-class.

Communication Quality

Papers report quantitative results transparently (particle counts, resolutions, accessions), but several are preprints with dense methods; no author-specific communication signals (reviews, textbooks) are attributable given identity ambiguity.

Author Novelty

The attributed corpus is highly novel — first near-atomic on-virion HA oligomer lattices, first resolved poxvirus portal pore, theory-guided de novo FDR calibration — though novelty is tied to unverified identity and 2025–2026 preprints lacking incoming citations.

Scientific Rigor

Recorded rigor is high: multi-tool pipelines (RELION/Dynamo/AlphaFold3/PHENIX), n=3 mutant replicates, public EMDB/PDB/GitHub/PRIDE deposits, and explicit falsification criteria. Deductions for AlphaFold3-dependent assignments, in vitro-only preparations, and database-label-dependent calibration.