Author Review — inspect what researchers actually reported
Aggregate an author's papers' raw data, methods, conflicts, and reproducibility cues.
Press Enter ↵ to lookup
Explore by Goal
"The more we learn about the world, and the deeper our learning, the more conscious, specific, and articulate will be our knowledge of what we do not know, our knowledge of our ignorance."
- Karl Popper
Quick Explanation
Copied
Lili Wang — scientific strength (evidence-based)
Across the provided record, the strongest signals are (i) mechanistic multi-omics/functional validation in cancer/immune contexts (e.g., CHST2 epigenetic activation in TNBC; METTL3→splicing-factor translation in CLL)
and (ii) quantitative platform work (spatial/epigenomic tissue mapping with multiplex imaging;
and deep learning for subcellular plant–pathogen interface quantification)
— paired with explicit limitations (pilot cohorts, heterogeneity, replication gaps) in the provided summaries.
TNBC epigenetic CHST2 driver framework:
CLL epitranscriptomic mechanistic axis:
Quantitative tissue/platform innovation in space:
Quantitative microscopy + DL for plant–microbe interfaces:
Long Explanation
BGPT Critical Author Review: Lili Wang
Date: April 13, 2026. Evidence used: only the papers explicitly summarized in the provided research data block (URLs/DOIs given there). Where the provided material is “correlative” or pilot-sized, I treat conclusions as hypothesis-generating rather than definitive.
1) Evidence-first visual evidence map
The provided record spans multiple biological subdomains (epigenetic cancer drivers; immune-oncology spatial architecture; CLL epitranscriptomics; plant–pathogen interface quantification; mechanotransduction biophysics; replication-stress signaling in plants). The scientific question I’m evaluating: does the author’s work demonstrate mechanistic grounding + quantitative rigor?
Figure A — TNBC epigenetic feature sizes (CHST2 paper)
Derived directly from the provided extracted data: TCGA BRCA TNBC vs non-TNBC counts and differentially methylated promoters (DMPs) split into hypomethylated vs hypermethylated.
Uses only the quantities provided: number of patients (n=4), claimed biomarker multiplex range (30–60), and detected niches (14).
Figure C — Plant-interface DL detection quality (HFinder)
Summarizes the provided haustoria test-set results: TP, FN, FP, total, and median IoU for NRC4-positive haustoria.
2) What seems scientifically strong
Mechanistic directionality is tested, not only observed. In the TNBC CHST2 work, the summary explicitly includes promoter-level methylation changes, an inverse methylation–expression relationship, and functional perturbations including a catalytic mutant where CHST2 catalytic activity is required for the full migration/invasion phenotype.
Multi-omics integration with functional validation at multiple molecular layers. The METTL3/CLL axis summary reports coordinated transcriptome + proteome + translatome/translation-relevant readouts, alongside genetic perturbation (METTL3 KO/knockdowns and catalytically informed manipulations) and in vivo xenograft validation.
Platform development aimed at scaling measurement. CmTSA is presented as an approach for multiplexed biomarker profiling on archival FFPE, coupled to AI segmentation and niche detection.
Quantitative, reproducible computational tooling for microscopy. HFinder reports explicit detection metrics (TP/FN/FP, median IoU) and provides model/data resources via Zenodo, while demonstrating cross-pathosystem learning.
3) Skeptical critique: major blind spots & fragilities (from the provided summaries)
Correlative-to-causal leaps are a recurring risk. Even when mechanisms are supported, multiple provided summaries explicitly label dependence on heterogeneous public datasets and potential batch effects, and they note limited in vivo or cohort size.
Translation fragility across models/species/tissue contexts. Where the provided summaries involve cross-species comparisons, model systems (cell lines, FFPE sections, mice) can preserve some signals while distorting others (e.g., regulatory landscapes differ).
Data availability and reanalysis risk. For some provided summaries, code/data are described as available “upon reasonable request” or largely via contacting authors. That increases the risk that key processing decisions are not easily independently audited.
4) Overall scientific score rationale (what this record suggests)
Based on the specific provided examples, Lili Wang’s scientific work appears strongest where there is (1) quantitative measurement (omics, multiplex imaging, or microscopy quantification), (2) explicit computational structure (pipelines, niche detection, DL segmentation), and (3) attempts at mechanistic grounding via perturbation experiments.
Confidence is moderate because several summaries explicitly flag pilot sizes, reliance on heterogeneous datasets, and limited external/independent replication within the provided material.
Build a summary table and Plotly charts from the provided extracted counts/metrics to compare study scales (cohort size, multiplex range, DMP splits, detection TP/FN/FP) and visualize evidence-strength signals.
Get emailed when your analysis is done!
We'll email you the results when your analysis is finished.
Hypothesis Graveyard
A “pure marker model” where CHST2-high merely tags EMT without being enzymatically required for invasive behavior—this is less favored because the provided summary includes catalytic mutant attenuation tied to migration/invasion effects.
A “dataset-only” explanation for CmTSA niche prognostic associations—i.e., that niches are clustering artifacts of preprocessing; this is less favored because the summary reports quenching, segmentation, and comparative clustering behavior (RNN vs KNN), though pilot size limits certainty.
Science Art
Science Movie
Make a narrated HD Science movie for this answer ($32 per minute)
Explore by Goal
Quick Explanation
Long Explanation
Top Data Sources
Ask a Follow-Up
Key Insight
Analysis 
Hypothesis Graveyard
Discussion