Infoblock alpha
q-bio/single-cell IB-arxiv-2602.17532-scfm-gene-auroc measured

For perturbation prediction in scGPT and Geneformer, trivial gene-level baselines outperformed attention and correlation edges (AUROC 0.81-0.88 versus 0.70).

Active Evidence state is recorded explicitly and may be revised by a later event.

Observed quantities

auroc_gene_lo 0.81 · auroc_gene_hi 0.88 · auroc_edges 0.7

01

Primary source

  1. arxiv · 2602.17532 · v1 · abstract
Title
Systematic Evaluation of Single-Cell Foundation Model Interpretability Reveals Attention Captures Co-Expression Rather Than Unique Regulatory Signal
Authors
Ihor Kendiukhov
Version
v1
Check
Support quote found word for word in the arXiv abstract v1, 2026-10-08
Supporting quote from the source
trivial gene-level baselines outperform both attention and correlation edges (AUROC 0.81-0.88 versus 0.70)

02

Boundary of this block

No additional public limitation was recorded.

03

Connections

Outgoing

No outgoing public edges.

Incoming

No incoming public edges.