Mardiyyah/TAPT_data_V2_split · Datasets at Hugging Face

PMCID stringclasses 24 values	Title stringclasses 24 values	Sentences stringlengths 2 40.7k
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scPoli model was saved for the downstream comparison.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each organoid sample, the same set of variable genes used in the primary tissue atlas (adult or fetal) was chosen, and the scPoli query was executed using identical parameters to those used in the primary tissue atlas training model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The UMAP embedding was transformed using the primary tissue atlas UMAP model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each cell, the system selected its 100 nearest neighbors from the HEOCA dataset.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The predicted tissue for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To compare and correlate cell states in primary tissue and organoid models, the miloR method was used to define and construct neighborhood graphs for each data source separately.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We computed the transcriptional similarity graph for the primary tissue reference using 30 nearest neighbors and the UMAP representation of latent representations of integrated primary tissue cells.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To compute the transcriptional similarity graph for the organoid reference, we used the 30 nearest neighbors and the UMAP representation of integrated embedding of organoid cells.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Single-cell organoid data were integrated using scPoli and 3,000 highly variable genes as described earlier.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We used the default parameters for all the remaining computational steps in building the neighborhood graphs.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We then used the R package scrabbitr to compute the correlation between each pair of neighborhoods in the primary tissue and organoid reference and to annotate the results at cell-type or tissue level.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The neighborhood correlations were computed using 3,000 highly variable genes that were found in the highly variable genes in the primary tissue single-cell reference atlases.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	This step results in two neighborhood correlation matrices: a primary tissue-correlation matrix in which each entry marks correlation of the expression profile of a given neighborhood in the primary tissue with the HEOCA, and an organoid-correlation matrix that stores the correlation of expression profiles in each neighborhood of the organoid atlas with the primary tissue atlas.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	This procedure was also repeated for each organoid derivation protocol, that is ASC-, FSC- and PSC-derived protocols.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To compare the correlation between cell states in the primary tissue and between organoid derivation protocols, we subtracted the primary tissue–neighborhood correlation matrices computed with respect to neighborhoods for each derivation protocol.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	This approach of comparing primary tissue and organoid by correlation of neighborhood graphs is more reliable than the alternative reference mapping strategy, because it removes the dependance of the reliability and accuracy of the conclusions to mapping uncertainty, and allows for computing correlation statistics on graphs that are constructed based on transcriptional similarity of cells in each data source.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For RNA velocity analysis of the HIOCA, we first excluded samples missing splicing information.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We then applied scVelo to generate a UMAP representation with stream trajectory visualization.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The velocity pseudotime, spanning from stem cells to enterocytes and colonocytes, has been rescaled to a range of 0 to 1.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We calculated and displayed the average expression of markers in specific bins.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The top 3,000 highly variable genes were subsetted for data integration.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To integrate all the cells, we applied the scPoli method with the same parameters used in the HEOCA atlas integration.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scPoli model was saved for the downstream comparison.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Lung organoid single-cell data curated from different studies was subsetted on top 3,000 highly variable genes for integration.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We applied scPoli to learn 30-dimensional latent representations of the cells, and 10-dimensional latent representations of the samples using a neural network with 2 hidden layers each of size 512.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The network was trained setting n_epochs=12, pretraining epochs to 10, eta=10, patience=20, lr_patience=13, lr_factor=0.1, alpha_epoch_anneal=100, reduced_lr=True and prototypical loss of the validation set as the early stopping criteria.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scPoli model was saved for the downstream comparison.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scRNA-seq data from both duodenum fetal and adult primary tissues were obtained from two research papers.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We focused on epithelial cells and subsetted them for analysis.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The top 3,000 highly variable genes were subsetted for data integration.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To integrate all the cells, we applied the scPoli method with the same parameters used in the HEOCA atlas integration.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scPoli model was saved for the downstream comparison.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each organoid sample, the same set of variable genes used in the primary tissue atlas was chosen, and the scPoli query was executed using identical parameters to those used in the primary tissue atlas training model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The UMAP embedding was transformed using the primary tissue atlas UMAP model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each cell, the system selected its 100 nearest neighbors from the primary tissue dataset.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The predicted cell type for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To identify DEGs in primary tissue stem cells and enterocytes, we subsetted these cell types and used a linear model to calculate the covariance between sample age and gene expression for each gene.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The top 100 genes with the highest coefficients were selected as DEGs.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The GSEApy method was then applied to identify the top GO-enriched terms associated with these genes.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To identify the heterogeneity of intestinal organoid stem cells and enterocytes, cells from the HIOCA were subsetted.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Integration was performed using the CSS method based on 1,000 highly variable genes across all cells.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Leiden clustering with a resolution of 0.1 was applied to identify subclusters.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The Wilcoxon rank-sum test was used to identify DEGs among subclusters, and GO enrichment analysis was conducted on the top 500 DEGs of each group using GSEApy.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scRNA-seq data from both duodenum fetal and adult primary tissues were obtained from two research papers.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The top 3,000 highly variable genes were subsetted for data integration.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To integrate all the cells, we applied the scPoli method with the same parameters used in the HEOCA atlas integration.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scPoli model was saved for the downstream comparison.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each organoid sample, the same set of variable genes used in the primary tissue atlas was chosen, and the scPoli query was executed using identical parameters to those used in the primary tissue atlas training model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The UMAP embedding was transformed using the primary tissue atlas UMAP model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each cell, the system selected its 100 nearest neighbors from the primary tissue dataset.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The predicted cell type for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Samples of scRNA-seq raw reads were mapped to the human genome, and counts of the matrix were obtained.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The same set of variable genes used in HEOCA was chosen, and the scPoli query was executed using identical parameters to those used in the HEOCA training model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The UMAP embedding was transformed using the HEOCA UMAP model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each cell, the system selected its 100 nearest neighbors from the HEOCA dataset.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The predicted cell type for the cell was determined by assigning it the cell type that was most frequently observed among its 100 nearest neighbors at the level 2 cell-type classification.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Similarly, the predicted tissue for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each cell in the organoid protocols, organoid perturbation, and disease samples, a matched HEOCA cell was reconstructed using the top ten kNN in HEOCA.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The mean expression of these ten neighbors was calculated to represent the expression profile of the matched sample reference in HEOCA.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	In addition, the mean kNN distance of these ten neighbors was used to represent the cell’s distance to the HEOCA.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To compare expression levels of the samples, the above-mentioned matched sample reference in HEOCA was identified.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The expression difference per gene for each cell pair was calculated based on the log-normalized expression values.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each gene, the variance over the calculated expression difference per cell pair was compared with the sum of squared expression differences normalized by the number of cell pairs.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	An F test was applied to test for differential expression for each gene.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	scRNA-seq reads for each sample were mapped to the human genome, and gene counts were generated using CellRanger.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	These counts served as input for sc2heoca, with default settings used to map all samples to HEOCA.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	During mapping, each cell was annotated with a level 2 cell type, and the distance to HEOCA was calculated.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Cell proportions were determined based on the mapping annotations.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For the raw integration of perturbation samples, three samples were merged, highly variable genes were identified using Scanpy with default settings, and the samples were integrated using the ComBat method.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The sc2heoca package with default settings was used to identify DEGs, and GO enrichment analysis was performed using GSEApy on the DEGs of each group.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	DEGs between each treatment sample and control sample were calculated using the Wilcoxon rank-sum test in Scanpy (v.1.9.3).
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The scIBD database was downloaded, and samples from healthy individuals, and patients with colitis, Crohn’s disease and ulcerative colitis were extracted.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Only epithelial cells were selected for downstream analysis.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Pseudo-bulk gene expression was calculated for each individual, and the DEGs identified in the previous step were subsetted.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The mean expression of these genes across patients was used to compare gene expression between inflammatory and viral response conditions.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The disease sample analysis is similar to the sample incorporation step.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The raw count matrices were downloaded from the original papers.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The same set of variable genes used in HEOCA was chosen, and the scPoli query was executed using identical parameters to those used in the HEOCA training model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The UMAP embedding was transformed using the HEOCA UMAP model.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each cell, the system selected its ten nearest neighbors from the HEOCA dataset.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The predicted cell type for the cell was determined by assigning it the cell type that was most frequently observed among its ten nearest neighbors at the level 2 cell-type classification.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The predicted tissue for the cell was assigned based on the tissue that was most frequently observed among its ten nearest neighbors.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each cell, the mean distance of its ten nearest neighbors was assigned as its mean distance to HEOCA.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	In the analysis of DEGs between colon cancer organoid colonocytes and bronchial COPD organoid basal cells, we performed separate subsetting for all colonocytes and basal cells.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	For each dataset, we isolated the top 3,000 highly variable genes.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We then integrated these subsets of cells using the bbknn method.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To cluster the two datasets, we applied the Leiden method with resolutions of 1 and 2 in two datasets.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The clusters predominantly associated with the disease were selected as disease state cells, while the remaining clusters were categorized as normal state cells.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We used D2C to score the expression levels of 2,395 drug target signatures in single cells from the human organoid cell atlas and the human lung cell atlas.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	D2C scores were scaled to mitigate scale differences between different datasets in the atlas.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We used the R package scDECAF (v.0.99.0) to select drug target signatures that exhibited global covariation in one or more cell types in HEOCA and HLCA primary tissue atlases.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The inputs to scDECAF were the scaled D2C z-scores and the cell embeddings from the atlases.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	The shrinkage operator in scDECAF was set to lambda = exp(−1.3) based on reconstruction error plots made available in the scDECAF package.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	We assigned a drug signature to a cell type if more than 50% of the cells from the cell type had a signature score above median across all cell types.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Multicellular drug target signatures were identified whether a drug signature was selected in at least two cell types.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	To assess druggability potential of organoid cell types, we computed the cosine similarity for cell-type pairs in organoid and primary tissue based on multicellular drug signatures identified in primary tissue and organoid models.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
PMC12081310	An integrated transcriptomic cell atlas of human endoderm-derived organoids.	Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41588-025-02182-6.
PMC12141184	Curcumin inhibits IFN-γ induced PD-L1 expression via reduction of STAT1 Phosphorylation in A549 non-small cell lung cancer cells.	Immune evasion in non-small cell lung cancer (NSCLC) is largely mediated by programmed death-ligand 1 (PD-L1), which is upregulated by interferon-gamma (IFN-γ)-induced STAT1 activation.
PMC12141184	Curcumin inhibits IFN-γ induced PD-L1 expression via reduction of STAT1 Phosphorylation in A549 non-small cell lung cancer cells.	Targeting this pathway may improve immunotherapy outcomes.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scPoli model was saved for the downstream comparison.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each organoid sample, the same set of variable genes used in the primary tissue atlas (adult or fetal) was chosen, and the scPoli query was executed using identical parameters to those used in the primary tissue atlas training model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The UMAP embedding was transformed using the primary tissue atlas UMAP model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each cell, the system selected its 100 nearest neighbors from the HEOCA dataset.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The predicted tissue for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To compare and correlate cell states in primary tissue and organoid models, the miloR method was used to define and construct neighborhood graphs for each data source separately.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We computed the transcriptional similarity graph for the primary tissue reference using 30 nearest neighbors and the UMAP representation of latent representations of integrated primary tissue cells.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To compute the transcriptional similarity graph for the organoid reference, we used the 30 nearest neighbors and the UMAP representation of integrated embedding of organoid cells.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Single-cell organoid data were integrated using scPoli and 3,000 highly variable genes as described earlier.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We used the default parameters for all the remaining computational steps in building the neighborhood graphs.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We then used the R package scrabbitr to compute the correlation between each pair of neighborhoods in the primary tissue and organoid reference and to annotate the results at cell-type or tissue level.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The neighborhood correlations were computed using 3,000 highly variable genes that were found in the highly variable genes in the primary tissue single-cell reference atlases.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

This step results in two neighborhood correlation matrices: a primary tissue-correlation matrix in which each entry marks correlation of the expression profile of a given neighborhood in the primary tissue with the HEOCA, and an organoid-correlation matrix that stores the correlation of expression profiles in each neighborhood of the organoid atlas with the primary tissue atlas.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

This procedure was also repeated for each organoid derivation protocol, that is ASC-, FSC- and PSC-derived protocols.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To compare the correlation between cell states in the primary tissue and between organoid derivation protocols, we subtracted the primary tissue–neighborhood correlation matrices computed with respect to neighborhoods for each derivation protocol.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

This approach of comparing primary tissue and organoid by correlation of neighborhood graphs is more reliable than the alternative reference mapping strategy, because it removes the dependance of the reliability and accuracy of the conclusions to mapping uncertainty, and allows for computing correlation statistics on graphs that are constructed based on transcriptional similarity of cells in each data source.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For RNA velocity analysis of the HIOCA, we first excluded samples missing splicing information.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We then applied scVelo to generate a UMAP representation with stream trajectory visualization.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The velocity pseudotime, spanning from stem cells to enterocytes and colonocytes, has been rescaled to a range of 0 to 1.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We calculated and displayed the average expression of markers in specific bins.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The top 3,000 highly variable genes were subsetted for data integration.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To integrate all the cells, we applied the scPoli method with the same parameters used in the HEOCA atlas integration.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scPoli model was saved for the downstream comparison.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Lung organoid single-cell data curated from different studies was subsetted on top 3,000 highly variable genes for integration.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We applied scPoli to learn 30-dimensional latent representations of the cells, and 10-dimensional latent representations of the samples using a neural network with 2 hidden layers each of size 512.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The network was trained setting n_epochs=12, pretraining epochs to 10, eta=10, patience=20, lr_patience=13, lr_factor=0.1, alpha_epoch_anneal=100, reduced_lr=True and prototypical loss of the validation set as the early stopping criteria.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scPoli model was saved for the downstream comparison.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scRNA-seq data from both duodenum fetal and adult primary tissues were obtained from two research papers.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We focused on epithelial cells and subsetted them for analysis.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The top 3,000 highly variable genes were subsetted for data integration.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To integrate all the cells, we applied the scPoli method with the same parameters used in the HEOCA atlas integration.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scPoli model was saved for the downstream comparison.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each organoid sample, the same set of variable genes used in the primary tissue atlas was chosen, and the scPoli query was executed using identical parameters to those used in the primary tissue atlas training model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The UMAP embedding was transformed using the primary tissue atlas UMAP model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each cell, the system selected its 100 nearest neighbors from the primary tissue dataset.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The predicted cell type for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To identify DEGs in primary tissue stem cells and enterocytes, we subsetted these cell types and used a linear model to calculate the covariance between sample age and gene expression for each gene.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The top 100 genes with the highest coefficients were selected as DEGs.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The GSEApy method was then applied to identify the top GO-enriched terms associated with these genes.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To identify the heterogeneity of intestinal organoid stem cells and enterocytes, cells from the HIOCA were subsetted.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Integration was performed using the CSS method based on 1,000 highly variable genes across all cells.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Leiden clustering with a resolution of 0.1 was applied to identify subclusters.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The Wilcoxon rank-sum test was used to identify DEGs among subclusters, and GO enrichment analysis was conducted on the top 500 DEGs of each group using GSEApy.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scRNA-seq data from both duodenum fetal and adult primary tissues were obtained from two research papers.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The top 3,000 highly variable genes were subsetted for data integration.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To integrate all the cells, we applied the scPoli method with the same parameters used in the HEOCA atlas integration.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scPoli model was saved for the downstream comparison.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each organoid sample, the same set of variable genes used in the primary tissue atlas was chosen, and the scPoli query was executed using identical parameters to those used in the primary tissue atlas training model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The UMAP embedding was transformed using the primary tissue atlas UMAP model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each cell, the system selected its 100 nearest neighbors from the primary tissue dataset.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The predicted cell type for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Samples of scRNA-seq raw reads were mapped to the human genome, and counts of the matrix were obtained.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The same set of variable genes used in HEOCA was chosen, and the scPoli query was executed using identical parameters to those used in the HEOCA training model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The UMAP embedding was transformed using the HEOCA UMAP model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each cell, the system selected its 100 nearest neighbors from the HEOCA dataset.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The predicted cell type for the cell was determined by assigning it the cell type that was most frequently observed among its 100 nearest neighbors at the level 2 cell-type classification.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Similarly, the predicted tissue for the cell was assigned based on the tissue that was most frequently observed among its 100 nearest neighbors.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each cell in the organoid protocols, organoid perturbation, and disease samples, a matched HEOCA cell was reconstructed using the top ten kNN in HEOCA.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The mean expression of these ten neighbors was calculated to represent the expression profile of the matched sample reference in HEOCA.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

In addition, the mean kNN distance of these ten neighbors was used to represent the cell’s distance to the HEOCA.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To compare expression levels of the samples, the above-mentioned matched sample reference in HEOCA was identified.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The expression difference per gene for each cell pair was calculated based on the log-normalized expression values.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each gene, the variance over the calculated expression difference per cell pair was compared with the sum of squared expression differences normalized by the number of cell pairs.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

An F test was applied to test for differential expression for each gene.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

scRNA-seq reads for each sample were mapped to the human genome, and gene counts were generated using CellRanger.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

These counts served as input for sc2heoca, with default settings used to map all samples to HEOCA.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

During mapping, each cell was annotated with a level 2 cell type, and the distance to HEOCA was calculated.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Cell proportions were determined based on the mapping annotations.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For the raw integration of perturbation samples, three samples were merged, highly variable genes were identified using Scanpy with default settings, and the samples were integrated using the ComBat method.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The sc2heoca package with default settings was used to identify DEGs, and GO enrichment analysis was performed using GSEApy on the DEGs of each group.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

DEGs between each treatment sample and control sample were calculated using the Wilcoxon rank-sum test in Scanpy (v.1.9.3).

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The scIBD database was downloaded, and samples from healthy individuals, and patients with colitis, Crohn’s disease and ulcerative colitis were extracted.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Only epithelial cells were selected for downstream analysis.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Pseudo-bulk gene expression was calculated for each individual, and the DEGs identified in the previous step were subsetted.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The mean expression of these genes across patients was used to compare gene expression between inflammatory and viral response conditions.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The disease sample analysis is similar to the sample incorporation step.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The raw count matrices were downloaded from the original papers.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The same set of variable genes used in HEOCA was chosen, and the scPoli query was executed using identical parameters to those used in the HEOCA training model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The UMAP embedding was transformed using the HEOCA UMAP model.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each cell, the system selected its ten nearest neighbors from the HEOCA dataset.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The predicted cell type for the cell was determined by assigning it the cell type that was most frequently observed among its ten nearest neighbors at the level 2 cell-type classification.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The predicted tissue for the cell was assigned based on the tissue that was most frequently observed among its ten nearest neighbors.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each cell, the mean distance of its ten nearest neighbors was assigned as its mean distance to HEOCA.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

In the analysis of DEGs between colon cancer organoid colonocytes and bronchial COPD organoid basal cells, we performed separate subsetting for all colonocytes and basal cells.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

For each dataset, we isolated the top 3,000 highly variable genes.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We then integrated these subsets of cells using the bbknn method.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To cluster the two datasets, we applied the Leiden method with resolutions of 1 and 2 in two datasets.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The clusters predominantly associated with the disease were selected as disease state cells, while the remaining clusters were categorized as normal state cells.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We used D2C to score the expression levels of 2,395 drug target signatures in single cells from the human organoid cell atlas and the human lung cell atlas.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

D2C scores were scaled to mitigate scale differences between different datasets in the atlas.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We used the R package scDECAF (v.0.99.0) to select drug target signatures that exhibited global covariation in one or more cell types in HEOCA and HLCA primary tissue atlases.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The inputs to scDECAF were the scaled D2C z-scores and the cell embeddings from the atlases.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

The shrinkage operator in scDECAF was set to lambda = exp(−1.3) based on reconstruction error plots made available in the scDECAF package.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

We assigned a drug signature to a cell type if more than 50% of the cells from the cell type had a signature score above median across all cell types.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Multicellular drug target signatures were identified whether a drug signature was selected in at least two cell types.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

To assess druggability potential of organoid cell types, we computed the cosine similarity for cell-type pairs in organoid and primary tissue based on multicellular drug signatures identified in primary tissue and organoid models.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

PMC12081310

An integrated transcriptomic cell atlas of human endoderm-derived organoids.

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41588-025-02182-6.

PMC12141184

Curcumin inhibits IFN-γ induced PD-L1 expression via reduction of STAT1 Phosphorylation in A549 non-small cell lung cancer cells.

Immune evasion in non-small cell lung cancer (NSCLC) is largely mediated by programmed death-ligand 1 (PD-L1), which is upregulated by interferon-gamma (IFN-γ)-induced STAT1 activation.

PMC12141184

Curcumin inhibits IFN-γ induced PD-L1 expression via reduction of STAT1 Phosphorylation in A549 non-small cell lung cancer cells.

Targeting this pathway may improve immunotherapy outcomes.