The DOI matches its Crossref record; 110 of the 110 references it cites were found in a registry or an index; none of the 12 sources its claims cite is checked for support yet.
scGPT: Towards Building a Foundation Model for Single-Cell Multi-omics Using Generative AI
Open accessAbstract
Generative pre-trained models have achieved remarkable success in various domains such as natural language processing and computer vision. Specifically, the combination of large-scale diverse datasets and pre-trained transformers has emerged as a promising approach for developing foundation models. Drawing parallels between linguistic constructs and cellular biology — where texts comprise words, similarly, cells are defined by genes — our study probes the applicability of foundation models to advance cellular biology and genetics research. Utilizing the burgeoning single-cell sequencing data, we have pioneered the construction of a foundation model for single-cell biology, scGPT, which is based on generative pre-trained transformer across a repository of over 33 million cells. Our findings illustrate that scGPT, a generative pre-trained transformer, effectively distills critical biological insights concerning genes and cells. Through the further adaptation of transfer learning, scGPT can be optimized to achieve superior performance across diverse downstream applications. This includes tasks such as cell-type annotation, multi-batch integration, multi-omic integration, genetic perturbation prediction, and gene network inference. The scGPT codebase is publicly available at https://github.com/bowang-lab/scGPT .
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
scGPT is a generative pretrained transformer foundation model for single-cell biology, pretrained on over 33 million human scRNA-seq cells and fine-tuned for cell-type annotation, integration, perturbation prediction, and gene network inference. It reports state-of-the-art or competitive performance across these tasks and identifies a scaling law where larger pretraining data improves downstream fine-tuning.
Key findings
- Pretraining used 33 million normal human scRNA-seq cells from CELLxGENE, spanning 51 organs/tissues and 441 studies; UMAP of a 10% subset showed cell types separated into distinct clusters despite over 400 studies.
- Cell-type annotation: scGPT achieved high precision (>0.8) for most hPancreas cell types except rare types, around 0.85 accuracy on multiple sclerosis, and ≥0.7 precision for most tumor-infiltrating myeloid cell types; it outperformed TOSICA and scBert on Accuracy, Precision, Recall, and MacroF1.
- Perturbation prediction: scGPT achieved the highest correlation on both Adamson and Norman datasets versus GEARS and CPA; in reverse perturbation prediction it identified 71.4% correct perturbations (5/7 test cases) within top 5 predictions, a 95.2% speedup over random tryouts (105.5 average attempts out of 210).
- Integration and GRN: PBMC 10K AvgBIO was 0.821, 5–10% higher than compared methods; ASAP PBMC AvgBATCH was 0.951 versus scMoMaT’s 0.916; scGPT uniquely identified 22 Reactome pathways, 14 immune-related, and attention analysis found all top 20 DDIT3-influenced genes and 19/20 BHLHE40-influenced genes as ChIP-Atlas targets.
Methods
- Design: two-stage generative pretraining and fine-tuning of a transformer that jointly learns cell and gene embeddings; a specialized attention mask enables generative prediction on non-sequential gene expression, with fine-tuning objectives including GEP, GEPC, ECS, DAR, and CLS.
- Data: pretraining on 33 million normal human scRNA-seq cells from CELLxGENE; downstream datasets included hPancreas, multiple sclerosis, myeloid, Adamson, Norman, COVID-19, PBMC 10K, Perirhinal Cortex, 10X Multiome PBMC, BMMC, ASAP PBMC, and Immune Human.
- Analysis: benchmarks against TOSICA, scBert, GEARS, CPA, scVI, Seurat, Harmony, scGLUE, and scMoMaT; metrics included Accuracy, Precision, Recall, MacroF1, Pearson correlation and corr(Δ), AvgBIO, AvgBATCH, Overall, and Reactome pathway enrichment.
Limitations
- Stated: the scaling law appeared to saturate after 3M pretraining cells; the authors hypothesize larger model sizes could improve performance at 33M and beyond but leave larger model sizes to future work.
- Stated: future work plans larger-scale, more diverse pretraining including multi-omic, spatial omics, diseased conditions, perturbation and temporal data, broader biological validation, and in-context instruction learning.
- Evident: this is a bioRxiv preprint not certified by peer review; results are computational benchmarks, and the text does not report prospective wet-lab validation of predictions.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
37 Own result · 15 Own interpretation · 13 Own method · 8 Not checked yet · 4 No basis given · 3 Sources not found
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“the fine-tuned scGPT consistently showcases performance gain for tasks such as integration376 and cell type annotation.”
§ 1 3 > 2.6 Analyzing the transfer learning of pretrained knowledge: the scaling370
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“scGPT achieves a higher overall batch correction287 score AvgBATCH of 0.951 compared to scMoMat (AvgBATCH = 0.916).”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“scGPT334 consistently demonstrates a significantly higher number of enriched pathways across all clustering335 resolutions.”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
boostedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“scGPT stands264 out as the only method that successfully generates a distinct cluster for CD8 Naive cells, while the265 other two methods fail to do so.”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The top performers in data integration tasks were models pretrained415 on whole-human, blood, and lung datasets”
§ 1 3 > 2.6 Analyzing the transfer learning of pretrained knowledge: the scaling370
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
In the paper
“scGPT identified the top 20 genes most influenced by353 DDIT3 transcription factor repression, which are all found to be signaling targets of DDIT3 in354 the Chip-Atlas database”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
What the sources say
- Not checked yetcites [56] 2022, ChIP-Atlas 2021 update: a data-mining suite for exploring epigenomic landscapes by fully integrating chip-seq, ATAC-seq and Bisulfite-seq data, In: Nucleic acids research
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Notably, scGPT specifically highlighted pathways related to the adaptive immune system, T-cell340 receptor (TCR) signaling, PD-1 signaling, and MHC II presentation, which were not captured by341 the co-expression network.”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Rests on an assumption: Adaptive immune populations exist in the fine-tuning datasets.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In the context of the Perirhinal Cortex dataset,246 scGPT remains competitive against all other methodologies”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Notably, scGPT also demonstrates a considerable performance244 for integrating PBMC 10K dataset even without fine-tuning”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
boosted- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“scGPT's biological288 conservation metrics also compare favorably to scMoMat's”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Probably not a substantial claim. 26
- It may state two separate claims as one. 17
- The quote only partly states this claim. 13
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“clustering resolution of 0.5, 54 sub-clusters were identified in the UMAP of predicted perturbation”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Both methods identi-337 fied 15 common pathways, including those associated with the cell cycle and the immune system.338 However, scGPT uniquely identified additional 22 pathways, out of which 14 were immune-related.”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“scMoMat produced285 two distinct clusters for each cell type corresponding to the first two and second two batches,286 indicating failure to learn a joint representation.”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
hedgeddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This superior integra-240 tion performance of scGPT is further supported by its high biological conservation score, with an241 AvgBIO score of 0.821, which is 5-10% higher than the compared methods.”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observed that a set of HLA class II genes was identified as group 1. Similarly, the CD3 genes327 involved in the T3 complex were identified as group 4, with the highest expression present in T328 cells.”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“scGPT presented more defined cluster structures compared269 to Seurat v4, with a 9% improvement in the AvgBIO score.”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observed the performance of fine-tuned models improved as the383 size of the pre-training data increased.”
§ 1 3 > 2.6 Analyzing the transfer learning of pretrained knowledge: the scaling370
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The quote only partly states this claim.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“source of genetic change within 5 attempts represents a 95.2% speedup.”
§ 1 3 > 2.3 scGPT predicts unseen genetic perturbation responses158
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: unclear — It is unclear why the paper says it holds.
It may state two separate claims as one.
In the paper
“The pre-trained scGPT model successfully identifies the group of315 genes (CD3E, D, and G) encoding the T3 complex for T-cell activation, as well as CD79A and316 CD79B for B-cell signaling, and CD8A and CD8B as co-receptors for HLA class I molecules”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
What the sources say
- Not checked yetcites [52] 1995, Immunobiology: The immune system in health and disease, In: Journal of Allergy and Clinical Immunology
boostedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“10X Multiome PBMC [47] scGPT (fine-tuned) 0.758 0.807 0.822 0.645 - - - -”
- Also stated in: § Supplementary Notes1264 · plain
“scGLUE [14] 0.747 0.815 0.806 0.619 - - - -”
- Also stated in: § Supplementary Notes1264 · plain
“Seurat v4 [46] 0.722 0.784 0.691 0.691 - - - -”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one. The quote only partly states this claim.
- Also stated in: § Supplementary Notes1264 · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“ASAP PBMC [49] scGPT (fine-tuned) 0.587 0.645 0.469 0.648 0.951 0.909 0.992 0.732”
- Also stated in: § Supplementary Notes1264 · plain
“scMoMat [15] 0.546 0.448 0.557 0.633 0.916 0.849 0.983 0.667”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one. The quote only partly states this claim.
- Also stated in: § Supplementary Notes1264 · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“scGPT (Zero-shot) Celltype, AvgBIO = 0.728”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
The quote only partly states this claim.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“BMMC [48] scGPT (fine-tuned) 0.697 0.783 0.725 0.582 0.871 0.834 0.908 0.766”
- Also stated in: § Supplementary Notes1264 · plain
“Seurat v4 [46] 0.600 0.737 0.470 0.594 0.678 0.685 0.671 0.631”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one. The quote only partly states this claim.
- Also stated in: § Supplementary Notes1264 · plain
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“From data1318 presented in table S5, we noticed a significant drop in the training time per epoch as well as in GPU1319 Memory usage throughout the fine-tuning process.”
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Specifically, the removal of zero-expressed genes1320 resulted in a substantial reduction in our model's maximum sequence length to approximately 401321 ∼ 60% of its initial length, effectively halving both the time spent on training epochs and peak1322 GPU Memory utilization.”
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Furthermore, by freezing the embedding layer, we achieved an additional1323 reduction in Peak GPU Memory usage by 1GB, a slight increase in training speed, while preserving1324 a comparable AvgBio score.”
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“These findings319 demonstrate scGPT's ability to generalize the knowledge learned during pre-training and extract320 information relevant to specific tissues or cell types after fine-tuning.”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
boosteddirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Overall, scGPT demon-273 strates superior cell-type clustering performance in the paired data setting and exhibits robustness274 across diverse biological conservation metrics benchmarked”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
boostedcorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
It may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“1,200 highly variable genes, 4,000 highly variable peaks, and all 216 protein features were used as”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
In the paper
“The evaluation of the integrated cell embeddings was performed using biological conservation”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
What the sources say
- Not checked yetcites Luecken et al. [53] 2022, Benchmarking atlas-level data integration in single-cell genomics, In: Nature methods
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The quote only partly states this claim.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“coefficient (corr) between the predicted gene expression and the ground-truth expression values.”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. It may state two separate claims as one. The quote only partly states this claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“making it particularly suitable for capturing interactions between genes.”
§ 4 Methods462 > 4.2 Cell and gene expression modelling by transformers513
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds.
In the paper
“outlined by Roohani, Huang, and Leskovec [38] in their benchmark.”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
What the sources say
- Not matchedcites Roohani, Huang, and Leskovec [38]
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
- Not matched
In the paper
“against the functional groups found in the original publication by Norman et al. [37]”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
What the sources say
- Not checked yetcites Norman et al. [37] 2019, Exploring genetic interaction manifolds constructed from rich single-cell phenotypes, In: Science
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
None of the sources cited for this claim could be matched to a paper in Crossref, so there is nothing to check it against.
In the paper
“we performed pathway enrich-331 ment analysis against the Reactome database [54] and identified high-confidence”
§ 1 3 > 2.5 scGPT uncovers gene regulatory networks for specific cell states and292
What the sources say
- Not matchedcites [54] Reactome. Reactome Pathway Database: Home. https://reactome.org/. 2022.
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
- Not matchedcites [55] Eric W Weisstein . “Bonferroni correction”. In: https://mathworld.wolfram.com/ (2004).
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds. The quote only partly states this claim.
- Not matched
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“While keeping the input feature set consistent, we used each method's”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The quote only partly states this claim.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“comparison since the method was not specifically designed to model protein data.”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
In the paper
“Additionally, we reported the batch correction metrics proposed by Luecken et al. [53] to assess”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
What the sources say
- Not checked yetcites Luecken et al. [53] 2022, Benchmarking atlas-level data integration in single-cell genomics, In: Nature methods
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“all methods were provided with the same number of 1,200 highly variable genes as input.”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“all models in the same settings as follows”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“As a sensitivity analysis, we reported the pathway hits for scGPT”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The quote only partly states this claim.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“the train-test split consists of three scenarios with increasing difficulty”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The quote only partly states this claim.
In the paper
“is a weighted sum of AvgBIO and AvgBATCH, consistent with the approach taken by Luecken”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
What the sources say
- Not checked yetcites Luecken et al. [53] 2022, Benchmarking atlas-level data integration in single-cell genomics, In: Nature methods
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“are split to ensure that test perturbations are not seen in training”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
In the paper
“Harmony and scVI are highlighted as the top-performing methods”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
What the sources say
- Not checked yetcites Luecken et al. [53] 2022, Benchmarking atlas-level data integration in single-cell genomics, In: Nature methods
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“Bonferroni correction [55] based on the total number of tests performed”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
What the sources say
- Not matchedcites [55] Eric W Weisstein . “Bonferroni correction”. In: https://mathworld.wolfram.com/ (2004).
Couldn’t be matched to a paper in Crossref, so it can’t be checked.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The quote only partly states this claim.
- Not matched
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Self-attention has been widely used to capture the co-occurrence patterns among tokens.”
§ 4 Methods462 > 4.3 Generative pre-training570
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The same 1,200 highly variable genes and 4,000 highly variable peaks were used”
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This setting serves as an example of out-of-distribution data.”
§ 4 Methods462 > 4.6 Datasets775
plainn = 9- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The difference between these two attention maps should highlight the”
§ 4 Methods462 > 4.5 Fine-tuning on downstream tasks701
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it. The paper may give no reason why it holds.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“The data has only one technical batch. Batch correction metrics are not applicable to this setting.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: fails — The paper’s conclusions may not depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The gene targets selected from the control attention map should reflect the basal”
§ 4 Methods462 > 4.5 Fine-tuning on downstream tasks701
plaincorrelational- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: fails — The paper may give no reason why it holds.
Probably not a substantial claim. The paper may give no reason why it holds. It may state two separate claims as one. The quote only partly states this claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“that the columns in the attention map indicate how much the gene of interest affects the other”
§ 4 Methods462 > 4.5 Fine-tuning on downstream tasks701
plaindirect- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
Probably not a substantial claim.
The paper relies on this without support.
The COVID-19 dataset comprises diverse cellular compositions, making it suitable for evaluating in-context pre-training effects.
“The COVID-19 dataset serves as an ideal task for this investigation, as it comprises diverse412 cellular compositions.”
The paper relies on this without support.
Adaptive immune populations exist in the fine-tuning datasets.
“This is concordant with the fact that adaptive immune populations342 exist in the fine-tuning datasets.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Empirical benchmarking studies have documented substantial performance degradation for biological foundation models under cross-platform transfer, preprocessing variability, and rare-population evaluation [20–24].”
From The next paradigm in bioinformatics: a review of multi-agent systems and foundational models for end-to-end scientific discovery (2026) · cites this paper as [20]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Compared to tasks such as cell annotation and batch correction, perturbation prediction offers experimentally generated model input/output with well-defined objectives, making it a more robust benchmark for assessing foundation models (15, 20, 21).”
From A systematic comparison of single-cell perturbation response prediction models (2026) · cites this paper as [15]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“We chose these models because they offer pretrained weights and have been trained using unsupervised objectives on extensive datasets (ca. 30M single-cell transcriptomes).”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“However, scGPT directly learns a cell embedding, which is integrated into its pretraining loss of predicting masked genes: scGPT first predicts a masked gene expression bin and a cell embedding from unmasked genes; then, in a second step, it further iteratively refines masked gene expression using the cell embedding predicted in the first step.”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Recent advances have incorporated graph-based methods and foundational models that take advantage of prior knowledge of gene expression relationships, resulting in improved generalizability (14–17).”
From A systematic comparison of single-cell perturbation response prediction models (2026) · cites this paper as [15]
Experimental — accuracy not yet measured
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Recent work suggests that architectural choices, such as generative versus contrastive pretraining objectives, introduce distinct trade-offs between representation flexibility, interpretability, and robustness to experimental variability [20–23, 36].”
From The next paradigm in bioinformatics: a review of multi-agent systems and foundational models for end-to-end scientific discovery (2026) · cites this paper as [20]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Both scGPT and Geneformer fine-tune their cell embeddings for cell type classification.”
From Assessing the limits of zero-shot foundation models in single-cell biology (2023) · cites this paper as [6]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“These models are pre-trained on vast amounts (> 10 M examples) of unlabelled scRNA-seq data, with the aim that such large datasets allow the models to capture general principles of gene regulation and signalling.”
From Benchmarking foundation cell models for post-perturbation RNA-seq prediction (2025) · cites this paper as [6]
The cited source says what the claim attributes to it.
In the citing paper
“Next, the binned values and the gene tokens (i.e., a unique index for each gene) are separately embedded and summed in the embedding space—jointly representing the gene and its binned expression.”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
The cited source says what the claim attributes to it.
In the citing paper
“Foundation models for single-cell and multi-omic biology, such as scGPT [20], Nicheformer [21], and EpiAgent [22, 23], produce high-quality continuous representations of biological entities that support cross-tissue transfer, perturbation inference, and regulatory feature learning.”
From The next paradigm in bioinformatics: a review of multi-agent systems and foundational models for end-to-end scientific discovery (2026) · cites this paper as [20]
The cited source says what the claim attributes to it.
In the citing paper
“We first examine recent foundational models that produce transferable representations across omic modalities, such as scGPT, Nicheformer, and EpiAgent, and discuss their architectural choices, training regimes, and interpretability constraints.”
From The next paradigm in bioinformatics: a review of multi-agent systems and foundational models for end-to-end scientific discovery (2026) · cites this paper as [20]
The cited source says what the claim attributes to it.
In the citing paper
“Single-cell foundation models and perturbation predictors support cellular-response forecasting [48-51].”
From RSI-Anything: Unlocking Recursive Self-Improvement with World Models (2026) · cites this paper as [48]
In the citing paper
“scGPT is available in several variants, each pretrained on multiple different datasets.”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
In the citing paper
“Finally, compared to Geneformer, scGPT has 3 the number of parameters, using 12 Transformer layers with 8 attention heads.”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
In the citing paper
“scGPT preprocesses each gene expression vector by independently binning values into 50 equidistant bins where the lowest bin is the lowest expression and the highest bin corresponds to the highest expression.”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
In the citing paper
“Like Geneformer, scGPT uses an MLM task.”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
In the citing paper
“To assess this, we employ metrics based on Average Silhouette Width (ASW) [25] and Average Bio (AvgBIO) scores [7].”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
In the citing paper
“To evaluate the performance of scGPT in its pretraining objective, we used the mean squared error (MSE), as used by the original authors for the model’s loss [7].”
From Zero-shot evaluation reveals limitations of single-cell foundation models (2025) · cites this paper as [7]
In the citing paper
“scGPT [14] focuses on processing these high-dimensional and sparse RNA data to capture complex relationships between cells, thereby supporting transfer learning across datasets and even across species.”
From Challenges in AI-driven Biomedical Multimodal Data Fusion and Analysis (2025) · cites this paper as [14]
In the citing paper
“Empirical benchmarking studies have shown that performance improvements observed in standard evaluation settings can decrease substantially when models are tested across different laboratories, sequencing platforms, or preprocessing pipelines [20–23].”
From The next paradigm in bioinformatics: a review of multi-agent systems and foundational models for end-to-end scientific discovery (2026) · cites this paper as [20]
In the citing paper
“We developed xTrimoGene, a scalable transformer-based model with both algorithmically efficient and engineering acceleration strategies 14 .”
From Large-scale foundation model on single-cell transcriptomics (2024) · cites this paper as [14]
In the citing paper
“With the help of self-attention mechanisms over the gene dimension, scGPT can encode complex interactions between perturbed and other genes.”
From A mini-review on perturbation modelling across single-cell omic modalities (2024) · cites this paper as [69]
In the citing paper
“This enables the model to learn from existing experimental data (i.e., Perturb-seq data) through few-shot learning and make accurate predictions of gene expressions following perturbation for unseen experiments (OOD).”
From A mini-review on perturbation modelling across single-cell omic modalities (2024) · cites this paper as [69]
In the citing paper
“In oncology, transformer-based methods support clinical text processing, whole-slide and radiologic image interpretation (vision transformers), and genomic sequence modeling [24,25,26,27].”
From Artificial Intelligence in Clinical Oncology: From Productivity Enhancement to Creative Discovery (2025) · cites this paper as [27]
In the citing paper
“To evaluate their models, most previous works-including scGPT and Geneformer-rely on fine-tuning to specialize task-specific models.”
From Assessing the limits of zero-shot foundation models in single-cell biology (2023) · cites this paper as [6]
In the citing paper
“We chose these models because they offer pretrained weights (whereas several other possible models did not have public weights at time of evaluation) and have been trained using unsupervised objectives on extensive datasets (ca. 30M single-cell transcriptomes).”
From Assessing the limits of zero-shot foundation models in single-cell biology (2023) · cites this paper as [6]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ Supplementary Notes1264
“AvgBATCH = 0.6 ∗AvgBIO + 0.4 ∗AvgBATCH”
§ Supplementary Notes1264
“AvgBATCH = (ASWbatch +GraphConn)/2”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 4 Methods462 > 4.7 Benchmarking Experiment Setup860
“batch correction and cell type clustering on four integration datasets: COVID-19 [13], PBMC 10K930 [44], and Perirhinal Cortex [45].”
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
“The evaluation was conducted on three integration236 datasets, namely the COVID-19 (18 batches) [13], PBMC 10K (2 batches) [44], and Perirhinal237 Cortex (2 batches) [45].”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 1 3 > 2.2 scGPT improves the precision of cell type annotation120
“fewer than 50 cells belong to the mast and MHC class II cell types out of the 106,000134 cells in the reference set.”
§ 4 Methods462 > 4.6 Datasets775
“The reference set contains 10,600 cells of 13 cell groups (alpha, beta, ductal,807 acinar, delta, PSC, PP, endothelial, macrophage, mast, epsilon, schwann, and t cell). The query808 set contains 4,218 cells of 11 cell groups (alpha, beta, ductal, PP, acinar, delta, PSC, endothelial,809 epsilon, mast, MHC class II).”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 1 3 > 2.2 scGPT improves the precision of cell type annotation120
“out of the 106,000134 cells in the reference set.”
§ 4 Methods462 > 4.6 Datasets775
“The reference set contains 10,600 cells of 13 cell groups”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
§ 1 3 > 2.4 scGPT enables accurate multi-batch and multi-omics integration231
“subgroup annotations (48 cell types).”
§ 4 Methods462 > 4.6 Datasets775
“annotations consist of 45 detailed immune cell subtypes.”
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“The original Perturb-seq study covered 236 one-gene or two-gene190 perturbations targeting 105 genes. However, considering all possible combinations of these tar-191 get genes, there are a total of 5,565 potential perturbations. This means that the experimental192 Perturb-seq only represents 5% of the entire perturbation space.”
§ 1 3 > 2.3 scGPT predicts unseen genetic perturbation responses158
Reported 5 · recomputed 4.240791
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2018) · arXiv · doi:10.48550/arxiv.1810.04805 · Cited by 73,173
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., et al. (2020) · arXiv · doi:10.48550/arxiv.2010.11929 · Cited by 47,472
Highly accurate protein structure prediction with AlphaFold
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., et al. (2021) · Nature · doi:10.1038/s41586-021-03819-2 · Cited by 43,764
Language Models are Few-Shot Learners
Brown, T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.14165 · Cited by 23,372 · cited 2×
“The generative pre-training used in OpenAI GPT3 [71] and GPT4 [19] employs a unified 585”
MethodRoBERTa: A Robustly Optimized BERT Pretraining Approach
Liu, Y., Ott, M., Goyal, N., Du, J., Joshi, M., Chen, D., Levy, O., Lewis, M., Zettlemoyer, L., & Stoyanov, V. (2019) · arXiv · doi:10.48550/arxiv.1907.11692 · Cited by 16,515
Integrated analysis of multimodal single-cell data
Hao, Y., Hao, S., Andersen-Nissen, E., Mauck, W. M., Zheng, S., Butler, A., Lee, M. J., Wilk, A. J., Darby, C., Zagar, M., et al. (2020) · doi:10.1101/2020.10.12.335331 · Cited by 15,508 · cited 5×
“In the paired data integration experiment, we benchmarked scGPT with scGLUE 951 [14] and Seurat v4 [46] on the 10X Multiome PBMC [47] dataset with RNA and ATAC-seq data as 952 the first example.”
MethodInfluentialComprehensive Integration of Single-Cell Data
Stuart, T., Butler, A., Hoffman, P., Hafemeister, C., Papalexi, E., Mauck, W. M., Hao, Y., Stoeckius, M., Smibert, P., & Satija, R. (2019) · Cell · doi:10.1016/j.cell.2019.05.031 · Cited by 15,453
Comprehensive integration of single-cell dataPreprint
Stuart, T., Butler, A. W., Hoffman, P. J., Hafemeister, C., Papalexi, E., Mauck, W., Hao, Y.-H., Stoeckius, M., Smibert, P., & Satija, R. (2018) · bioRxiv · doi:10.1101/460147 · Cited by 13,858
Fast, sensitive and accurate integration of single-cell data with Harmony
Korsunsky, I., Millard, N., Fan, J., Slowikowski, K., Zhang, F., Wei, K., Baglaenko, Y., Brenner, M., Loh, P.-R., & Raychaudhuri, S. (2019) · Nature Methods · doi:10.1038/s41592-019-0619-0 · Cited by 10,488 · cited 5×
“In our benchmarking experiments, we compared scGPT with three popular integration methods: scVI [41], Seurat [42], and Harmony [43].”
MethodInfluential
Research works printed but not linked: 6
- 2019, In: Chinese Computational Linguistics: 18th China National Conference, CCL 2019, Kunming, China, October 18–20, 2019, Proceedings, 18, 194
- National Library of Medicine. Bhlhe40 basic helix-loop-helix family, member e40 [Mus mus-culus (house mouse)]. https://www.ncbi.nlm.nih.gov/gene/20893. Online; accessed 29 May 2023. 2023.
- 2019, Language models are unsupervised multitask learners, In: OpenAI blog, 1, 9
- 10.1186/S13059-020-02000-8/FIGURES/6 · DOI
- 2013, Auto-encoding variational bayes, In: arXiv
- 2021, Joint probabilistic modeling of single-cell multi-omic data with to-talVI, In: Nature methods, 18, 272
Cited by
Sorted by citations
Large-scale foundation model on single-cell transcriptomics
Hao, M., Gong, J., Zeng, X., Liu, C., Guo, Y., Cheng, X., Wang, T., Ma, J., Zhang, X., & Song, L. (2024) · Nature Methods · doi:10.1038/s41592-024-02305-7 · Cited by 628
Program, C. C. S., Abdulla, S., Aevermann, B., Assis, P., Badajoz, S., Bell, S. M., Bezzi, E., Cakir, B., Chaffer, J., Chambers, S., et al. (2025) · Nucleic Acids Research · doi:10.1093/nar/gkae1142 · Cited by 399
How to build the virtual cell with artificial intelligence: Priorities and opportunities
Bunne, C., Roohani, Y., Rosen, Y., Gupta, A., Zhang, X., Roed, M., Alexandrov, T., AlQuraishi, M., Brennan, P., Burkhardt, D. B., et al. (2024) · Cell · doi:10.1016/j.cell.2024.11.015 · Cited by 393
Self-citationProfiling cell identity and tissue architecture with single-cell and spatial transcriptomics
Gulati, G. S., D'Silva, J. P., Liu, Y., Wang, L., & Newman, A. M. (2024) · Nature Reviews Molecular Cell Biology · doi:10.1038/s41580-024-00768-2 · Cited by 347
Deep learning-based approaches for multi-omics data integration and analysis
Ballard, J. L., Wang, Z., Li, W., Shen, L., & Long, Q. (2024) · BioData Mining · doi:10.1186/s13040-024-00391-z · Cited by 301
“Recently, Cui et al. [52] proposed scGPT, a foundation model for single-cell omics data.”
BackgroundA Vision-Language Foundation Model for Precision Oncology
Xiang, J., Wang, X., Zhang, X., Xi, Y., Eweje, F., Chen, Y., Li, Y., Bergstrom, C., Gopaulchan, M., Kim, T., et al. (2025) · Nature · doi:10.1038/s41586-024-08378-w · Cited by 256
Generative artificial intelligence empowers digital twins in drug discovery and clinical trials
Bordukova, M., Makarov, N., Rodriguez-Esteban, R., Schmich, F., & Menden, M. P. (2024) · Expert Opinion on Drug Discovery · doi:10.1080/17460441.2023.2273839 · Cited by 235
Baião, A. R., Cai, Z., Poulos, R. C., Robinson, P. J., Reddel, R. R., Zhong, Q., Vinga, S., & Gonçalves, E. (2025) · Briefings in Bioinformatics · doi:10.1093/bib/bbaf355 · Cited by 274
The diversification of methods for studying cell–cell interactions and communication
Armingol, E., Baghdassarian, H., & Lewis, N. E. (2024) · Nature Reviews Genetics · doi:10.1038/s41576-023-00685-8 · Cited by 229
“Although these technologies have been reviewed in detail 21–26 , here we highlight examples that are pertinent to gaining biological insights about CCIs, and could help refine the aforementioned computational tools and validate their predictions, improving the fidelity of CCI inference.”
Advances in molecular pathology and therapy of non-small cell lung cancer
Huang, Q., Li, Y., Huang, Y., Wu, J., Bao, W., Xue, C., Li, X., Dong, S., Dong, Z., & Hu, S. (2025) · Signal Transduction and Targeted Therapy · doi:10.1038/s41392-025-02243-6 · Cited by 197
Wu, Y., & Xie, L. (2025) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2024.12.030 · Cited by 220
Decoding the tumor microenvironment with spatial technologies
Walsh, L. A., & Quail, D. F. (2023) · Nature Immunology · doi:10.1038/s41590-023-01678-9 · Cited by 199
Transformers in single-cell omics: a review and new perspectives
Szałata, A., Hrovatin, K., Becker, S., Tejada-Lapuerta, A., Cui, H., Wang, B., & Theis, F. J. (2024) · Nature Methods · doi:10.1038/s41592-024-02353-z · Cited by 189
Self-citationThe Human Cell Atlas from a cell census to a unified foundation model
Rood, J., Wynne, S., Robson, L., Hupalowska, A., Randell, J., Teichmann, S. A., & Regev, A. (2024) · Nature · doi:10.1038/s41586-024-08338-4 · Cited by 173
Hadi, M. U., Al-Tashi, Q., Qureshi, R., Shah, A., Muneer, A., Irfan, M., Zafar, A., Shaikh, M. B., Akhtar, N., Hassan, S. Z., et al. (2023) · doi:10.36227/techrxiv.23589741.v4 · Cited by 193
Universal Cell Embeddings: A Foundation Model for Cell BiologyPreprint
Rosen, Y., Roohani, Y., Agrawal, A., Samotorcan, L., Quake, S. R., & Leskovec, J. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.11.28.568918 · Cited by 165
“In particular, we compare against Geneformer [28] and scGPT [29].”
Ahlmann-Eltze, C., Huber, W., & Anders, S. (2025) · Nature Methods · doi:10.1038/s41592-025-02772-6 · Cited by 238
Yang, X., Liu, G., Feng, G., Bu, D., Wang, P., Jiang, J., Chen, S., Yang, Q., Zhang, Y., Man, Z., et al. (2023) · doi:10.1101/2023.09.26.559542 · Cited by 148
Yang, X., Liu, G., Feng, G., Bu, D., Wang, P., Jiang, J., Chen, S., Yang, Q., Miao, H., Zhang, Y., et al. (2024) · Cell Research · doi:10.1038/s41422-024-01034-y · Cited by 148
“To test whether the similarity between homologs was derived from the prior knowledge or the self-supervised pre-training, we compared the cosine similarity of homologous genes in various scenarios, including pre-trained GeneCompass with prior knowledge, Fig.”
Self-citationProgram, C. S.-C. B., Abdulla, S., Aevermann, B., Assis, P., Badajoz, S., Bell, S. M., Bezzi, E., Cakir, B., Chaffer, J., Chambers, S., et al. (2023) · doi:10.1101/2023.10.30.563174 · Cited by 142
Generative-AI-Driven Human Digital Twin in IoT Healthcare: A Comprehensive Survey
Chen, J., Shi, Y., Yi, C., Du, H., Kang, J., & Niyato, D. T. (2024) · IEEE Internet of Things Journal · doi:10.1109/jiot.2024.3421918 · Cited by 168
“The model further leveraged attention mechanism to capture intricate gene-to-gene interactions at the single-cell level [52].”
BackgroundA cell atlas foundation model for scalable search of similar human cells
Heimberg, G., Kuo, T. C.-Y., DePianto, D. J., Salem, O., Heigl, T., Diamant, N., Scalia, G., Biancalani, T., Turley, S. J., Rock, J. R., et al. (2024) · Nature · doi:10.1038/s41586-024-08411-y · Cited by 204
Rood, J. E., Hupalowska, A., & Regev, A. (2024) · Cell · doi:10.1016/j.cell.2024.07.035 · Cited by 139
Towards multimodal foundation models in molecular cell biology
Cui, H., Tejada-Lapuerta, A., Brbić, M., Sáez-Rodríguez, J., Cristea, S., Goodarzi, H., Lotfollahi, M., Theis, F. J., & Wang, B. (2025) · Nature · doi:10.1038/s41586-025-08710-y · Cited by 151
“Recently, models such as scGPT 45 , CellOracle 74 , Geneformer 46 , CellOT 75 , CPA 38 , chemcCPA 76 and GEARS 39 have shown initial success in perturbing learned cell embeddings to predict resulting expression profiles.”
InfluentialSelf-citationA foundation model of transcription across human cell types
(2025) · doi:10.1038/s41586-024-08391-z · Cited by 130
Foundation Model for Advancing Healthcare: Challenges, Opportunities and Future Directions
He, Y., Huang, F., Jiang, X., Nie, Y., Wang, M., Wang, J., & Chen, H. (2024) · IEEE Reviews in Biomedical Engineering · doi:10.1109/rbme.2024.3496744 · Cited by 162
Relmasira, S. C., Lai, Y. C., & Donaldson, J. P. (2023) · Sustainability · doi:10.3390/su151813595 · Cited by 180
Large Scale Foundation Model on Single-cell Transcriptomics
Hao, M., Gong, J., Zeng, X., Liu, C., Guo, Y., Cheng, X., Wang, T., Ma, J., Song, L., & Zhang, X. (2023) · doi:10.1101/2023.05.29.542705 · Cited by 639
Foundation models in bioinformatics
Guo, F., Guan, R., Li, Y., Liu, Q., Wang, X., Yang, C., & Wang, J. (2025) · National Science Review · doi:10.1093/nsr/nwaf028 · Cited by 127
A visual–omics foundation model to bridge histopathology with spatial transcriptomics
Chen, W., Zhang, P., Tran, T. N., Xiao, Y., Li, S., Shah, V. V., Cheng, H., Brannan, K. W., Youker, K., Lai, L., et al. (2025) · Nature Methods · doi:10.1038/s41592-025-02707-1 · Cited by 118
“Unlike single-cell foundation models like scGPT 57 , Geneformer 26 and scFoundation 27 , our approach models omics data as text, effectively bridging molecular and visual modalities.”
A foundational large language model for edible plant genomes
Mendoza‐Revilla, J., Trop, E., Gonzalez, L., Roller, M., Dalla-Torre, H., de Almeida, B. P., Richard, G., Caton, J., Carranza, N. L., Skwark, M. J., et al. (2024) · Communications Biology · doi:10.1038/s42003-024-06465-2 · Cited by 129
“We have provided the predicted impact of all possible single-nucleotide substitutions across all enhancer and promoter sequences for future research purposes.”
BackgroundGenePT: A Simple But Effective Foundation Model for Genes and Cells Built From ChatGPTPreprint
Chen, Y. T., & Zou, J. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.10.16.562533 · Cited by 106
“Our validation process follows that of Cui et al. [2] and consists of the following steps: 1. constructing gene networks based on the cosine similarities among the highly variable genes; 2. applying unsupervised Louvain clustering [38] to derive gene programs; and 3. qualitatively comparing the…”
InfluentialLe, J., Dian, Y., Zhao, D., Guo, Z., Luo, Z., Chen, X., Zeng, F., & Deng, G. (2025) · Molecular Cancer · doi:10.1186/s12943-025-02426-3 · Cited by 118
Zero-shot evaluation reveals limitations of single-cell foundation models
Kedzierska, K. Z., Crawford, L., Amini, A. P., & Lu, A. X. (2025) · Genome Biology · doi:10.1186/s13059-025-03574-x · Cited by 109
“To assess this, we employ metrics based on Average Silhouette Width (ASW) [25] and Average Bio (AvgBIO) scores [7].”
InfluentialGhosh, S., Zhao, X., Alim, M., Brudno, M., & Bhat, M. (2024) · Gut · doi:10.1136/gutjnl-2023-331740 · Cited by 110
Transformers and genome language models
Consens, M. E., Dufault, C., Wainberg, M., Forster, D., Karimzadeh, M., Goodarzi, H., Theis, F. J., Moses, A. M., & Wang, B. (2025) · Nature Machine Intelligence · doi:10.1038/s42256-025-01007-9 · Cited by 170
“While this Review focuses on DNA-based genome language models, single-cell RNA-seq language models follow many similar principles 40,41,132–136 .”
Self-citationExploring the role of large language models in the scientific method: from hypothesis to discovery
Zhang, Y., Khan, S. A., Mahmud, A., Yang, H., Lavin, A., Levin, M. G., Frey, J. G., Dunnmon, J., Evans, J., Bundy, A., et al. (2025) · npj Artificial Intelligence · doi:10.1038/s44387-025-00019-5 · Cited by 160
A Comprehensive Survey of Foundation Models in Medicine
Khan, W., Leem, S., See, K. B., Wong, J. K., Zhang, S., & Fang, R. (2025) · IEEE Reviews in Biomedical Engineering · doi:10.1109/rbme.2025.3531360 · Cited by 137
“5fromdataof 33,000genesobtainedfromGeneformer[91]andscGPT[92].”
InfluentialQiu, X., Li, H., Steeg, G. V., & Godzik, A. (2024) · Biomolecules · doi:10.3390/biom14030339 · Cited by 103
“In early 2020, Exscientia reported the initiation of Phase 1 clinical trials for DSP-1181, a compound designed to address obsessive-compulsive disorder.”
BackgroundThe Large Language Models on Biomedical Data Analysis: A Survey
Lan, W., Tang, Z., Liu, M., Chen, Q., Peng, W., Chen, Y. P., & Pan, Y. F. (2025) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2025.3530794 · Cited by 100
“In particular, scBERT demonstrates superiority in robustness and its ability to detect novel cell types. scGPT [76] is a single-cell multi-omics language model based on generative pre-training.”
CellFM: a large-scale foundation model pre-trained on transcriptomics of 100 million human cells
Zeng, Y., Xie, J., Shangguan, N., Wei, Z., Li, W., Su, Y., Yang, S., Zhang, C., Zhang, J., Fang, N., et al. (2025) · Nature Communications · doi:10.1038/s41467-025-59926-5 · Cited by 100
Democratizing protein language models with parameter-efficient fine-tuning
Sledzieski, S., Kshirsagar, M., Baek, M., Dodhia, R., Lavista Ferres, J., & Berger, B. (2024) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2405840121 · Cited by 86
Predicting cellular responses to perturbation across diverse contexts with State
Adduri, A. K., Gautam, D., Bevilacqua, B., Naghipourfar, M., Imran, A., Shah, R., Teyssier, N., Verma, R., Carpenter, C., Eraslan, B., et al. (2026) · Cell · doi:10.1016/j.cell.2026.07.052 · Cited by 100
Nicheformer: a foundation model for single-cell and spatial omics
Tejada-Lapuerta, A., Schaar, A. C., Gutgesell, R., Palla, G., Halle, L., Minaeva, M., Vornholz, L., Dony, L., Drummer, F., Richter, T., et al. (2025) · Nature Methods · doi:10.1038/s41592-025-02814-z · Cited by 100
Ge, S., Sun, S., Xu, H., Cheng, Q. S., & Ren, Z. (2025) · Briefings in Bioinformatics · doi:10.1093/bib/bbaf136 · Cited by 97
Epigenomic heterogeneity as a source of tumour evolution
Laisné, M., Lupien, M., & Vallot, C. (2024) · Nature reviews. Cancer · doi:10.1038/s41568-024-00757-9 · Cited by 93
The future of rapid and automated single-cell data analysis using reference mapping
Lotfollahi, M., Hao, Y., Theis, F. J., & Satija, R. (2024) · Cell · doi:10.1016/j.cell.2024.03.009 · Cited by 81
Benchmarking algorithms for single-cell multi-omics prediction and integration
Hu, Y., Wan, S., Luo, Y., Li, Y., Wu, T., Deng, W., Jiang, C., Jiang, S., Zhang, Y., Liu, N., et al. (2024) · Nature Methods · doi:10.1038/s41592-024-02429-w · Cited by 79
Language models for biological research: a primer
Simon, E. P., Swanson, K., & Zou, J. (2024) · Nature Methods · doi:10.1038/s41592-024-02354-y · Cited by 85
Hallmarks of artificial intelligence contributions to precision oncology
Chang, T., Park, S., Schäffer, A. A., Jiang, P., & Ruppin, E. (2025) · Nature Cancer · doi:10.1038/s43018-025-00917-2 · Cited by 83
Digital twins in healthcare: a comprehensive review and future directions
Rudsari, H. K., Tseng, B., Zhu, H., Song, L., Gu, C., Roy, A., Irajizad, E., Butner, J. D., Long, J. P., & Do, K.-A. (2025) · Frontiers in Digital Health · doi:10.3389/fdgth.2025.1633539 · Cited by 105
Partners, C. M. A., Bào, Y., Zhang, Z., Zhao, W., Xiao, J., Song, S., He, S., Zhang, G., Li, Y., Zhao, G., et al. (2025) · Nucleic Acids Research · doi:10.1093/nar/gkaf1172 · Cited by 100
Optimal transport for single-cell and spatial omics
Bunne, C., Schiebinger, G., Krause, A., Regev, A., & Cuturi, M. (2024) · Nature Reviews Methods Primers · doi:10.1038/s43586-024-00334-2 · Cited by 71
Challenges in AI-driven Biomedical Multimodal Data Fusion and Analysis
Liu, J., Cen, X., Yi, C., Wang, F.-A., Ding, J., Cheng, J., Wu, Q., Gai, B., Zhou, Y., He, R., et al. (2025) · Genomics, Proteomics & Bioinformatics · doi:10.1093/gpbjnl/qzaf011 · Cited by 75
Ahlmann-Eltze, C., Huber, W., & Anders, S. (2024) · doi:10.1101/2024.09.16.613342 · Cited by 268
Advances and applications in single-cell and spatial genomics
Wang, J., Ye, F., Chai, H., Jiang, Y., Wang, T., Ran, X., Xia, Q., Xu, Z., Fu, Y., Zhang, G., et al. (2024) · Science China Life Sciences · doi:10.1007/s11427-024-2770-x · Cited by 70
Nicheformer: A Foundation Model for Single-Cell and Spatial OmicsPreprint
Schaar, A. C., Tejada-Lapuerta, A., Palla, G., Gutgesell, R. M., Halle, L., Minaeva, M., Vornholz, L., Dony, L., Drummer, F., Bahrami, M., et al. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.04.15.589472 · Cited by 66
Harnessing artificial intelligence to fill global shortfalls in biodiversity knowledge
Pollock, L. J., Kitzes, J. A., Beery, S., Gaynor, K. M., Jarzyna, M. A., Aodha, O. M., Meyer, B. E., Rolnick, D., Taylor, G. W., Tuia, D., et al. (2025) · Nature Reviews Biodiversity · doi:10.1038/s44358-025-00022-3 · Cited by 82
Considerations for building and using integrated single-cell atlases
Hrovatin, K., Sikkema, L., Shitov, V. A., Heimberg, G., Shulman, M., Oliver, A. J., Mueller, M. F., Ibarra, I. L., Wang, H., Ramírez-Suástegui, C., et al. (2024) · Nature Methods · doi:10.1038/s41592-024-02532-y · Cited by 72
Integrated single-cell atlas of human atherosclerotic plaques
Traeuble, K., Munz, M., Pauli, J., Sachs, N., Vafadarnejad, E., Carrillo‐Roa, T., Mäegdefessel, L., Kastner, P., & Heinig, M. (2025) · Nature Communications · doi:10.1038/s41467-025-63202-x · Cited by 72
Cell2Sentence: Teaching Large Language Models the Language of BiologyPreprint
Levine, D. M., Rizvi, S. A., Lévy, S., Pallikkavaliyaveetil, N., Zhang, D., Chen, X., Ghadermarzi, S., Wu, R., Zheng, Z., Vrkic, I., et al. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.09.11.557287 · Cited by 68
“This formatting enables LLMs to process and interpret this information (see Figures 1 and 2) while also maintaining the richness and complexity in single-cell data (Figures 3 and 4).”
MethodInfluentialEnablers and challenges of spatial omics, a melting pot of technologies
Alexandrov, T., Sáez-Rodríguez, J., & Saka, S. K. (2023) · Molecular Systems Biology · doi:10.15252/msb.202110571 · Cited by 65
Evaluating the Utilities of Foundation Models in Single-cell Data Analysis
Liu, T., Li, K., Wang, Y., Li, H., & Zhao, H. (2023) · doi:10.1101/2023.09.08.555192 · Cited by 62
Schäfer, P. S. L., Dimitrov, D., Villablanca, E. J., & Sáez-Rodríguez, J. (2024) · Nature Immunology · doi:10.1038/s41590-024-01768-2 · Cited by 71
“Deep learning models that are trained on millions of scRNA-seq profiles can learn intricate relationships between genes, surpassing the capabilities of classical coexpression analysis 104,105 .”
Evaluating the Utilities of Foundation Models in Single‐Cell Data Analysis
Liu, T., Li, K., Wang, Y., Li, H., & Zhao, H. (2026) · Advanced Science · doi:10.1002/advs.202514490 · Cited by 91
Assessing the limits of zero-shot foundation models in single-cell biology
Kedzierska, K. Z., Crawford, L., Amini, A. P., & Lu, A. X. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.10.16.561085 · Cited by 61
“In this study, we assessed the zero-shot performance of two proposed foundation models in single-cell biology: Geneformer [5] and scGPT [6] - representative examples in a rapidly evolving field.”
BackgroundMethodInfluentialProgress and opportunities of foundation models in bioinformatics
Li, Q., Hu, Z., Wang, Y., Li, L., Fan, Y., King, I., Jia, G., Wang, S., Song, L., & Li, Y. (2024) · Briefings in Bioinformatics · doi:10.1093/bib/bbae548 · Cited by 72
“Therefore, customized fine-tuning can correct sequencing batches from multiple datasets [42].”
BackgroundMethodInfluentialFoundation models and intelligent decision-making: Progress, challenges, and perspectives
Huang, J., Xu, Y., Wang, Q., Wang, Q., Liang, X., Wang, F., Zhang, Z., Wei, W., Zhang, B., Huang, L., et al. (2025) · The Innovation · doi:10.1016/j.xinn.2025.100948 · Cited by 87
Saunders, R. A., Allen, W. E., Pan, X., Sandhu, J., Lu, J., Lau, T. K., Smolyar, K., Sullivan, Z. A., Dulac, C. G., Weissman, J. S., et al. (2025) · Cell · doi:10.1016/j.cell.2025.05.022 · Cited by 82
A Multimodal Knowledge-enhanced Whole-slide Pathology Foundation Model
Xu, Y., Wang, Y., Zhou, F., Ma, J., Jin, C., Yang, S., Li, J., Zhang, Z., Zhao, C., Zhou, H., et al. (2024) · arXiv · doi:10.48550/arxiv.2407.15362 · Cited by 60
scTab: Scaling cross-tissue single-cell annotation models
Fischer, F., Fischer, D. S., Mukhin, R. S., Isaev, A. V., Biederstedt, E., Villani, A., & Theis, F. J. (2024) · Nature Communications · doi:10.1038/s41467-024-51059-5 · Cited by 71
A multimodal knowledge-enhanced whole-slide pathology foundation model
Xu, Y., Wang, Y., Zhou, F., Ma, J., Jin, C., Yang, S., Li, J., Zhang, Z., Zhao, C., Zhou, H., et al. (2025) · Nature Communications · doi:10.1038/s41467-025-66220-x · Cited by 118
Deep Learning for Genomics: From Early Neural Nets to Modern Large Language Models
Yue, T., Wang, Y., Zhang, L., Gu, C., Xue, H., Wang, W., Lyu, Q., & Dun, Y. (2023) · International Journal of Molecular Sciences · doi:10.3390/ijms242115858 · Cited by 192
Artificial General Intelligence for Medical Imaging Analysis
Li, X., Zhao, L., Zhang, L., Wu, Z., Liu, Z., Jiang, H., Cao, C., Xu, S., Li, Y., Dai, H., et al. (2024) · IEEE Reviews in Biomedical Engineering · doi:10.1109/rbme.2024.3493775 · Cited by 91
Evaluating generalizability of artificial intelligence models for molecular datasets
Ektefaie, Y., Shen, A. N., Bykova, D., Marin, M. G., Žitnik, M., & Farhat, M. R. (2024) · Nature Machine Intelligence · doi:10.1038/s42256-024-00931-6 · Cited by 69
“https://doi.org/10.1038/s42256-024-00931-6”
InfluentialLarge language models in plant biology
Lam, H. Y. I., Ong, X. E., & Mutwil, M. (2024) · Trends in Plant Science · doi:10.1016/j.tplants.2024.04.013 · Cited by 70
Simple and effective embedding model for single-cell biology built from ChatGPT
Chen, Y. T., & Zou, J. (2024) · Nature Biomedical Engineering · doi:10.1038/s41551-024-01284-6 · Cited by 205
“The patterns of average gene expressions for identified gene programmes in different cell types are congruent with those previously identified in Cui et al. 3 .”
InfluentialData science opportunities of large language models for neuroscience and biomedicine
Bzdok, D., Thieme, A., Levkovskyy, O., Wren, P. B., Ray, T. S., & Reddy, S. (2024) · Neuron · doi:10.1016/j.neuron.2024.01.016 · Cited by 65
TMO-Net: an explainable pretrained multi-omics model for multi-task learning in oncology
Wang, F.-A., Zhuang, Z., Gao, F., He, R., Zhang, S., Wang, L., Liu, J., & Li, Y. (2024) · Genome Biology · doi:10.1186/s13059-024-03293-9 · Cited by 51
A mini-review on perturbation modelling across single-cell omic modalities
Gavriilidis, G. I., Vasileiou, V., Orfanou, A., Ishaque, N., & Psomopoulos, F. (2024) · Computational and Structural Biotechnology Journal · doi:10.1016/j.csbj.2024.04.058 · Cited by 54
A machine learning approach to leveraging electronic health records for enhanced omics analysis
Mataraso, S. J., Espinosa, C. A., Seong, D., Reincke, S., Berson, E., Reiss, J., Kim, Y., Ghanem, M., Shu, C.-H., James, T., et al. (2025) · Nature Machine Intelligence · doi:10.1038/s42256-024-00974-9 · Cited by 53
Large language models for drug discovery and development
Zheng, Y., Koh, H. Y., Ju, J., Yang, M., May, L. T., Webb, G. I., Li, L., Pan, S., & Church, G. M. (2025) · Patterns · doi:10.1016/j.patter.2025.101346 · Cited by 121
Viñas Torné, R., Wiatrak, M., Piran, Z., Fan, S., Jiang, L., Teichmann, S. A., Nitzan, M., & Brbić, M. (2026) · Nature Biotechnology · doi:10.1038/s41587-025-02777-8 · Cited by 58
A multi-center study on the adaptability of a shared foundation model for electronic health records
Guo, L., Fries, J., Steinberg, E., Fleming, S. L., Morse, K., Aftandilian, C. C., Posada, J., Shah, N. H., & Sung, L. (2024) · npj Digital Medicine · doi:10.1038/s41746-024-01166-w · Cited by 67
New horizons at the interface of artificial intelligence and translational cancer research
Yates, J., & Allen, E. M. V. (2025) · Cancer Cell · doi:10.1016/j.ccell.2025.03.018 · Cited by 58
Single-cell foundation models: bringing artificial intelligence into cell biology
Baek, S., Song, K., & Lee, I. (2025) · Experimental & Molecular Medicine · doi:10.1038/s12276-025-01547-5 · Cited by 46
Large‐Language‐Model‐Based AI Agent for Organic Semiconductor Device Research
Zhang, Q., Hu, Y., Yan, J., Zhang, H., Xie, X., Zhu, J., Li, H., Niu, X., Li, L., Sun, Y., et al. (2024) · Advanced Materials · doi:10.1002/adma.202405163 · Cited by 53
scPRINT: pre-training on 50 million cells allows robust gene network predictions
Kalfon, J., Samaran, J., Gabriel Peyré, A., & Cantini, L. (2025) · Nature Communications · doi:10.1038/s41467-025-58699-1 · Cited by 76
scGPT-spatial: Continual Pretraining of Single-Cell Foundation Model for Spatial TranscriptomicsPreprint
Wang, C. X., Cui, H., Zhang, A., Xie, R., Goodarzi, H., & Wang, B. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.02.05.636714 · Cited by 61
Self-citationKarimzadeh, M., Momen-Roknabadi, A., Cavazos, T. B., Fang, Y., Chen, N.-C., Multhaup, M., Yen, J. L., Ku, J., Wang, J., Zhao, X., et al. (2024) · Nature Communications · doi:10.1038/s41467-024-53851-9 · Cited by 55
Benchmarking DNA foundation models for genomic and genetic tasks
Feng, H., Wu, L., Zhao, B., Huff, C. D., Zhang, J. J., Wu, J., Lin, L., Wei, P., & Wu, C. (2025) · Nature Communications · doi:10.1038/s41467-025-65823-8 · Cited by 83
Generative Artificial Intelligence in Anatomic Pathology
Brodsky, V., Ullah, E., Bychkov, A., Song, A. H., Walk, E. E., Louis, P., Rasool, G., Singh, R. S., Mahmood, F., Bui, M. M., et al. (2025) · Archives of Pathology & Laboratory Medicine · doi:10.5858/arpa.2024-0215-ra · Cited by 48
Liu, X. F., Tao, Y., Cai, Z., Bao, P., Ma, H., Li, K., Li, M., Zhu, Y., & Lu, Z. J. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae316 · Cited by 54
scDiffusion: conditional generation of high-quality single-cell data using diffusion model
Luo, E., Hao, M., Wei, L., & Zhang, X. (2024) · Bioinformatics · doi:10.1093/bioinformatics/btae518 · Cited by 66
“Furthermore, GAN requires careful designing and tuning of model architectures, as well as optimization tricks, to achieve a stable training process (Saxena and Cao 2021, Yang et al. 2022), hindering the smooth application on new datasets and the generation of data under certain conditions.”
Benchmarking foundation cell models for post-perturbation RNA-seq prediction
Csendes, G., Sanz, G., Szalay, K. Z., & Szalai, B. (2025) · BMC Genomics · doi:10.1186/s12864-025-11600-2 · Cited by 47
“To address these challenges, several Transformer-based, foundation cell models have emerged recently [4, 6, 7].”
Zhou, Y., Dai, Q., Xu, Y., Wu, S., Cheng, M., & Zhao, B. T. (2025) · npj Precision Oncology · doi:10.1038/s41698-025-01082-6 · Cited by 50
Novae: a graph-based foundation model for spatial transcriptomics data
Blampey, Q., Benkirane, H., Bercovici, N., Mulder, K., Gessain, G., Ginhoux, F., André, F., & Cournède, P.-H. (2025) · Nature Methods · doi:10.1038/s41592-025-02899-6 · Cited by 62
Unlocking the power of spatial omics with AI
Coleman, K., Schroeder, A. R., & Li, M. (2024) · Nature Methods · doi:10.1038/s41592-024-02363-x · Cited by 46
Artificial intelligence and machine learning applications for cultured meat
Todhunter, M. E., Jubair, S., Verma, R., Saqe, R., Shen, K., & Duffy, B. (2024) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2024.1424012 · Cited by 51
Large language model for knowledge synthesis and AI-enhanced biomanufacturing
Li, W., Mao, Z., Xiao, Z., Liao, X., Koffas, M. A. G., Chen, Y., Ma, H., & Tang, Y. (2025) · Trends in biotechnology · doi:10.1016/j.tibtech.2025.02.008 · Cited by 49
Metadata
- Authors
- Cui, H., Wang, C., Maan, H., Pang, K., Luo, F., & Wang, B.
- Affiliations
- University Health Network (CA)
- University of Toronto (CA)
- Corresponding author
- Wang, Bo
- Venue
- Nature Methods
- Publisher
- openRxiv
- Published
- 2023-05-01
- Volume
- 21
- Issue
- 8
- Pages
- 1470-1480
- ISSN
- 1548-7091 (ISSN-L)
- OpenAlex
- W4392168151
- Cited by
- 1,694
- References
- 98
- Citations by year
- 2026: 918 · 2025: 521 · 2024: 234 · 2023: 19
- FWCI
- 186.38
- Type
- journal article
- Language
- en
- Open access
- green
- doi.org
- Licence
- cc-by-nc-nd
- Locations
- Nature Methods (publishedVersion)
- bioRxiv (Cold Spring Harbor Laboratory) (submittedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- Funders
- University Health Network
- Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada (DGECR-2020-00294, RGPIN-2020-06189)
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Single-cell and spatial transcriptomics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Bioinformatics and Genomic Networks
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Gene Regulatory Network Analysis
- Keywords
- scGPT, single-cell multi-omics, foundation models, generative pre-trained transformer, single-cell RNA sequencing, cell type annotation, multi-omics integration, perturbation response prediction, gene network inference, transfer learning, single-cell biology
- MeSH terms
- Multiomics, Algorithms, Artificial Intelligence, Humans, Computational Biology, Gene Regulatory Networks, Single-Cell Analysis
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-11-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
