Unverified50% confidenceFactExact time
GeneFlow is a framework that unifies flow-matching generative modeling with structured multimodal fusion for predicting gene expression from tissue images in spatial transcriptomics.
1
Sources
50%
Confidence
Long-term
Relevance
11/16/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedGeneFlow uses a histology tokenizer and a gene tokenizer to encode tissue morphology information and cross-gene interactions within spatial spots, respectively.74% similarUnverifiedGeneFlow directly predicts spatial gene expression rather than an intermediate velocity field, and does not run the full generative trajectory at inference.72% similarUnverifiedCurrent genomic foundation models treat the genome as a flat string and do not explicitly model the stochastic dynamics of cohesin- and CTCF-driven loop extrusion, the biophysical principles of phase-separated condensate formation, or the cell-type-specific combinatorial grammar of transcription factors.70% similarUnverifiedIn single-cell multi-omics data, FlowAlign can better align cell representations across omics modalities, aiding downstream cell type annotation and trajectory inference.68% similarUnverifiedGeneFlow is expected to promote the widespread application of spatial transcriptomics in clinical and research settings.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1174168API
curl https://kongchang.com/api/v1/knowledge/claims/1174168MCP
get_claim(id=1174168)