STCGAN: a novel cycle-consistent generative adversarial network for spatial transcriptomics cellular deconvolution.

{"sci_version": "1.2", "schema": "https://corpus.bntrg.co.uk/sci/1.0/sci.schema.json", "id": "pmid:39715685", "kind": "science-record", "revision": 2, "content_hash": "sha256:2b36f3f68c3effe94411d9832d796c012c9c60fa778d5b6c45fb4e4889bd3be8", "created_at": "2026-11-01T20:30:02Z", "updated_at": "2026-11-01T21:34:36Z", "generator": {"name": "corpus-assay-engine", "run_id": "reading-sci-2026-11-01"}, "paper": {"state": "complete", "title": "STCGAN: a novel cycle-consistent generative adversarial network for spatial transcriptomics cellular deconvolution.", "doi": "10.1093/bib/bbae670", "pmid": "39715685", "authors": ["Bo Wang", "Yahui Long", "Yuting Bai", "Jiawei Luo", "Chee Keong Kwoh"], "journal": "Briefings in bioinformatics", "year": 2024, "abstract": "MOTIVATION: Spatial transcriptomics (ST) technologies have revolutionized our ability to map gene expression patterns within native tissue context, providing unprecedented insights into tissue architecture and cellular heterogeneity. However, accurately deconvolving cell-type compositions from ST spots remains challenging due to the sparse and averaged nature of ST data, which is essential for accurately depicting tissue architecture. While numerous computational methods have been developed for cell-type deconvolution and spatial distribution reconstruction, most fail to capture tissue complexity at the single-cell level, thereby limiting their applicability in practical scenarios.\nRESULTS: To this end, we propose a novel cycle-consistent generative adversarial network named STCGAN for cellular deconvolution in spatial transcriptomic. STCGAN first employs a cycle-consistent generative adversarial network (CGAN) to pre-train on ST data, ensuring that both the mapping from ST data to latent space and its reverse mapping are consistent, capturing complex spatial gene expression patterns and learning robust latent representations. Based on the learned representation, STCGAN then optimizes a trainable cell-to-spot mapping matrix to integrate scRNA-seq data with ST data, accurately estimating cellular composition within each capture spot and effectively reconstructing the spatial distribution of cells across the tissue. To further enhance deconvolution accuracy, we incorporate spatial-aware regularization that ensures accurate cellular distribution reconstruction within the spatial context. Benchmarking against seven state-of-the-art methods on five simulated and real datasets from various tissues, STCGAN consistently delivers superior cell-type deconvolution performance.\nAVAILABILITY: The code of STCGAN can be downloaded from https://github.com/cs-wangbo/STCGAN and all the mentioned datasets are available on Zenodo at https://zenodo.org/doi/10.5281/zenodo.10799113.", "claims": [{"id": "claim-method", "statement": "For performance evaluation, all experiments are repeated 10 times, and root mean square error (RMSE) and Jensen–Shannon Divergence (JSD) are used to evaluate the cellular deconvolution performance on simulated datasets.", "kind": "methodological", "target": null, "evidence": [{"asset": "source:pmid:39715685", "locator": "sections[9] \"Experiment settings\" paragraph 3", "quote": "For performance evaluation, all experiments are repeated 10 times, and root mean square error (RMSE) and Jensen–Shannon Divergence (JSD) are used to evaluate the cellular deconvolution performance on simulated datasets."}], "state": "complete"}, {"id": "claim-data-access", "statement": "For the seqFISH+ dataset, both the ST data and the scRNA-seq data are obtained from (https://github.com/CaiGroup/seqFISH-PLUS). The MERFISH dataset is obtained from (https://datadryad.org/stash/dataset/doi:10.5061/dryad.8t8s248/), and the scRNA-seq data is obtained from (https://github.com/rdong08/spatialDWLS_dataset/tree/main/datasets). The DLPFC dataset is obtained from (http://research.libd.org/spatialLIBD/), and the scRNA-seq data is obtained from (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144136). The mouse brain tissue dataset is collected from (https://mouse.brain-map.org/static/atlas), and the scRNA-seq reference is acquired from (https://portal.brain-map.org/atlases-and-data/rnaseq/mouse-whole-cortex-and-hippocampus-10x). The human breast cancer is collected from (https://www.10xgenomics.com/resources/datasets/human-breast-cancer-block-a-section-1-1-standard-1-1-0), and the scRNA-seq data is obtained from (https://www.immunesinglecell.org/).", "kind": "methodological", "target": null, "evidence": [{"asset": "source:pmid:39715685", "locator": "sections[20] \"Data availability\" paragraph 2", "quote": "For the seqFISH+ dataset, both the ST data and the scRNA-seq data are obtained from (https://github.com/CaiGroup/seqFISH-PLUS). The MERFISH dataset is obtained from (https://datadryad.org/stash/dataset/doi:10.5061/dryad.8t8s248/), and the scRNA-seq data is obtained from (https://github.com/rdong08/spatialDWLS_dataset/tree/main/datasets). The DLPFC dataset is obtained from (http://research.libd.org/spatialLIBD/), and the scRNA-seq data is obtained from (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144136). The mouse brain tissue dataset is collected from (https://mouse.brain-map.org/static/atlas), and the scRNA-seq reference is acquired from (https://portal.brain-map.org/atlases-and-data/rnaseq/mouse-whole-cortex-and-hippocampus-10x). The human breast cancer is collected from (https://www.10xgenomics.com/resources/datasets/human-breast-cancer-block-a-section-1-1-standard-1-1-0), and the scRNA-seq data is obtained from (https://www.immunesinglecell.org/)."}], "state": "complete"}], "gaps": []}, "experiment": {"state": "complete", "objective": "", "steps": [{"id": "read_published_estimate", "order": 1, "name": "Read the published estimate from the source", "kind": "other", "state": "complete", "blocked_on": null, "description": "The claim, method and evidence span published with the source article.", "duration_hours": null, "parameters": {}, "inputs": ["source:pmid:39715685"], "outputs": ["the published estimate and its evidence"], "instruments": [], "reagents": []}, {"id": "obtain_source_data", "order": 2, "name": "Obtain the source data from the named repository", "kind": "data-preparation", "state": "complete", "blocked_on": null, "description": "The GEO series accession resolves and its sample accessions are listed. This lane did not download the dataset or read its licence.", "duration_hours": null, "parameters": {}, "inputs": ["data:geo:GSE144136"], "outputs": ["the source data with its repository provenance"], "instruments": [], "reagents": []}, {"id": "run_analysis", "order": 3, "name": "Run the published analysis", "kind": "analysis", "state": "complete", "blocked_on": null, "description": "Apply the paper's statistical method to the resolved series and compare the result with the published target.", "duration_hours": null, "parameters": {}, "inputs": ["data:geo:GSE144136"], "outputs": ["the replicated result and its comparison"], "instruments": [], "reagents": []}], "materials": [], "instruments": [], "reagents": [], "conditions": {}, "gaps": []}, "data": {"state": "partial", "assets": [{"id": "source:pmid:39715685", "role": "manuscript", "format": "xml", "media_type": "application/xml", "uri": "https://doi.org/10.1093/bib/bbae670", "sha256": "sha256:8010ce9b4e0fcd7bbd47e7e53bc42972594323d15af0bffc624403c73c76440e", "bytes": 67674, "access": "open", "state": "resolved", "reader": null, "reader_status": "reader_required", "identifier": "10.1093/bib/bbae670", "version": null, "note": "Full article fetched from PMC; the checksum covers the rendered section body, not the original XML bytes.", "structure": {"pmcid": "PMC11666287", "source": "europe_pmc_fulltext", "coverage": "whole_body_without_references", "text_hash_scope": "rendered_section_body"}}, {"id": "data:geo:GSE144136", "role": "raw", "format": "geo-series", "media_type": "application/vnd.ncbi.geo-series+json", "uri": "https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144136", "sha256": null, "bytes": null, "access": "open", "state": "resolved", "reader": null, "reader_status": "reader_required", "identifier": "GSE144136", "version": null, "note": "For the seqFISH+ dataset, both the ST data and the scRNA-seq data are obtained from (https://github.com/CaiGroup/seqFISH-PLUS). The MERFISH dataset is obtained from (https://datadryad.org/stash/dataset/doi:10.5061/dryad.8t8s248/), and the scRNA-seq data is obtained from (https://github.com/rdong08/spatialDWLS_dataset/tree/main/datasets). The DLPFC dataset is obtained from (http://research.libd.org/spatialLIBD/), and the scRNA-seq data is obtained from (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144136). The mouse brain tissue dataset is collected from (https://mouse.brain-map.org/static/atlas), and the scRNA-seq reference is acquired from (https://portal.brain-map.org/atlases-and-data/rnaseq/mouse-whole-cortex-and-hippocampus-10x). The human breast cancer is collected from (https://www.10xgenomics.com/resources/datasets/human-breast-cancer-block-a-section-1-1-standard-1-1-0), and the scRNA-seq data is obtained from (https://www.immunesinglecell.org/).", "structure": {"repository": "GEO", "access_route": "GEO accession URL", "access_statement_verbatim": "For the seqFISH+ dataset, both the ST data and the scRNA-seq data are obtained from (https://github.com/CaiGroup/seqFISH-PLUS). The MERFISH dataset is obtained from (https://datadryad.org/stash/dataset/doi:10.5061/dryad.8t8s248/), and the scRNA-seq data is obtained from (https://github.com/rdong08/spatialDWLS_dataset/tree/main/datasets). The DLPFC dataset is obtained from (http://research.libd.org/spatialLIBD/), and the scRNA-seq data is obtained from (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144136). The mouse brain tissue dataset is collected from (https://mouse.brain-map.org/static/atlas), and the scRNA-seq reference is acquired from (https://portal.brain-map.org/atlases-and-data/rnaseq/mouse-whole-cortex-and-hippocampus-10x). The human breast cancer is collected from (https://www.10xgenomics.com/resources/datasets/human-breast-cancer-block-a-section-1-1-standard-1-1-0), and the scRNA-seq data is obtained from (https://www.immunesinglecell.org/).", "access_statement_kind": "data_availability_statement", "sample_accessions": ["GSM4281974", "GSM4281975", "GSM4281976", "GSM4281977", "GSM4281978", "GSM4281979", "GSM4281980", "GSM4281981", "GSM4281982", "GSM4281983", "GSM4281984", "GSM4281985", "GSM4281986", "GSM4281987", "GSM4281988", "GSM4281989", "GSM4281990", "GSM4281991", "GSM4281992", "GSM4281993", "GSM4281994", "GSM4281995", "GSM4281996", "GSM4281997", "GSM4281998", "GSM4281999", "GSM4282000", "GSM4282001", "GSM4282002", "GSM4282003", "GSM4282004", "GSM4282005", "GSM4282006", "GSM4282007"], "sample_count": 34, "group_sizes": null, "group_sizes_basis": null, "platform": "20301", "taxon": "Homo sapiens", "title": "Single-nucleus RNA-seq in the post-mortem brain in major depressive disorder", "resolution_checked_at": "2026-11-01T20:16:08Z", "resolution_check": "identifier_resolution_only_no_download_no_licence_read", "downloaded": false, "licence_read": false}}], "gaps": ["asset:data:geo:GSE144136 not downloaded", "asset:data:geo:GSE144136 licence not read"]}, "analysis": {"state": "partial", "methods": [{"id": "method-1", "family": null, "specific": null, "label": "root mean square error (RMSE) and Jensen–Shannon Divergence (JSD)", "dsl_operations": [], "missing_operations": [], "expressible": null, "parameters": {}, "data_shapes": [], "evidence": {"score": 1.0, "band": "verbatim", "quote": "For performance evaluation, all experiments are repeated 10 times, and root mean square error (RMSE) and Jensen–Shannon Divergence (JSD) are used to evaluate the cellular deconvolution performance on simulated datasets."}}], "targets": [], "gaps": ["method.reader_named_unmapped"]}, "capability": {"state": "complete", "requirements": [{"capability_id": "capability:computational_analysis", "label": "Computational analysis platform", "assay_families": ["meta_analysis", "machine_learning", "survival_analysis", "statistical_regression", "pca", "molecular_docking", "network_analysis", "differential_expression", "gwas"], "availability_state": "owned", "tier_unlocked": "R1", "candidate": "Automated Assays Laravel application and simulated adapters"}], "gaps": []}, "readiness": {"state": "blocked", "tier": "R4", "blocking_axis": "method", "blocked_on": "method:not_determinable", "axes": {"method": {"tier": "R4", "expressible": false, "required_operations": [], "missing_operations": [], "basis": "The reader quoted the method the paper names (root mean square error (RMSE) and Jensen–Shannon Divergence (JSD)), but the corpus's method extractor could not map it onto a DSL operation, so the method is published but not expressible."}, "capability": {"tier": "R1", "capability_id": "capability:computational_analysis", "availability_state": "owned", "basis": "The computational analysis is owned. The physical bench the paper used is not, so a physical validation cannot be run here."}, "data": {"tier": "R1", "reachable": true, "accessions": ["GSE144136"], "resolved_accessions": ["GSE144136"], "statements": ["For the seqFISH+ dataset, both the ST data and the scRNA-seq data are obtained from (https://github.com/CaiGroup/seqFISH-PLUS). The MERFISH dataset is obtained from (https://datadryad.org/stash/dataset/doi:10.5061/dryad.8t8s248/), and the scRNA-seq data is obtained from (https://github.com/rdong08/spatialDWLS_dataset/tree/main/datasets). The DLPFC dataset is obtained from (http://research.libd.org/spatialLIBD/), and the scRNA-seq data is obtained from (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE144136). The mouse brain tissue dataset is collected from (https://mouse.brain-map.org/static/atlas), and the scRNA-seq reference is acquired from (https://portal.brain-map.org/atlases-and-data/rnaseq/mouse-whole-cortex-and-hippocampus-10x). The human breast cancer is collected from (https://www.10xgenomics.com/resources/datasets/human-breast-cancer-block-a-section-1-1-standard-1-1-0), and the scRNA-seq data is obtained from (https://www.immunesinglecell.org/)."], "basis": "The paper names GSE144136 in NCBI Gene Expression Omnibus and the accession resolves in the repository. No dataset was downloaded and no licence was read."}}, "can_run": {"dsl_operations": [], "assets": ["source:pmid:39715685", "data:geo:GSE144136"], "steps": ["run_analysis"]}, "must_build": {"dsl_operations": [], "converters": [], "readers": [], "capabilities": [], "data_gaps": ["data:geo:GSE144136"]}, "gaps": ["method:not_determinable"]}, "review": {"state": "partial", "machine_critique": {"kind": "extraction_review", "reader": {"model": "MiniMax-M3.1-Flash", "provider": "minimax", "provider_family": "minimax", "read_at": "2026-11-01T19:43:07Z", "independence": "single_reader"}, "evidence_check": {"method": "verbatim", "availability": "verbatim"}, "calls": 5, "usage": {"prompt_tokens": 18577, "completion_tokens": 3208, "cached_tokens": 256}}, "human_critique": null, "stages": [{"id": "extraction", "state": "complete", "reader": "MiniMax-M3.1-Flash", "provider": "minimax"}], "gaps": ["review.independent_reader", "review.machine_critique", "review.human_critique"]}, "execution": {"state": "absent", "status": "not_run", "mode": null, "runs": [], "gaps": ["execution.not_run"]}, "provenance": {"entities": [{"id": "pmid:39715685", "type": "paper", "label": "pmid:39715685"}, {"id": "source:pmid:39715685", "type": "manuscript", "label": "PMID 39715685 full text"}, {"id": "data:geo:GSE144136", "type": "dataset", "label": "GEO GSE144136"}], "activities": [{"id": "activity-reading-sci-2026-11-01", "type": "classification", "used": [{"id": "pmid:39715685", "type": "paper", "label": "pmid:39715685"}, {"id": "source:pmid:39715685", "type": "manuscript", "label": "PMID 39715685 full text"}, {"id": "data:geo:GSE144136", "type": "dataset", "label": "GEO GSE144136"}], "agent": "agent:corpus-assay-engine"}, {"id": "activity-read-39715685", "type": "extraction", "used": ["source:pmid:39715685"], "agent": "agent:minimax-reader"}, {"id": "activity-resolve-GSE144136", "type": "accession_resolution", "used": ["data:geo:GSE144136"], "agent": "agent:corpus-assay-engine"}], "agents": [{"id": "agent:corpus-assay-engine", "type": "software", "label": "Corpus assay engine"}, {"id": "agent:minimax-reader", "type": "software", "label": "MiniMax literature reader"}], "source_run_id": "reading-sci-2026-11-01", "dsl_catalogue_fingerprint": "343fb30756ba6ec3", "capability_table_fingerprint": "2cb3e42f047d2ecf", "snapshot_id": "ba34b49490d9bfaf"}, "lineage": {"parent_id": "pmid:39715685", "parent_hash": "sha256:655b9bf8a72e6ec65f4960880c21f5bc7b28f198d05449c4f50ed09cf1d2c0cd", "fork_of": null, "generation": 1}, "gaps": ["asset:data:geo:GSE144136 licence not read", "asset:data:geo:GSE144136 not downloaded", "execution.not_run", "method.reader_named_unmapped", "review.human_critique", "review.independent_reader", "review.machine_critique"], "assays": [{"role": "data_generation", "definition_id": "scrna_seq", "definition_version": "99e1195c7d6f", "definition_digest": null, "required_operations": [], "support_state": "unsupported", "source_hashes": {"assay_lexicon": "99e1195c7d6f"}, "confidence": 0.99, "evidence": ["ble cell-to-spot mapping matrix to integrate scRNA-seq data with ST data, accurately estimating cel", "omprises two primary steps: pre-training and scRNA-seq mapping. In the pre-training stage, a CGAN i", "derlying data distribution (Fig. 1A). In the scRNA-seq mapping stage, we learn a trainable cell-to-", "learn a trainable cell-to-spot matrix to map scRNA-seq data to the predicted ST expression matrix \\"], "order_basis": "paper_text_order:1328", "id": "assay:1"}, {"role": "data_generation", "definition_id": "animal_model", "definition_version": "99e1195c7d6f", "definition_digest": null, "required_operations": [], "support_state": "partial", "source_hashes": {"assay_lexicon": "99e1195c7d6f"}, "confidence": 0.99, "evidence": ["on. STCGAN enables horizontal Integration of mouse brain samples The previous discussion has pr", "ated analysis of horizontal samples from the mouse brain anterior and posterior [27, 31]. We fi", "of STCGAN across the anterior and posterior mouse brain samples (Fig. 4A, S4–S5). Spatial scat", "horizontally integrates cell types from the mouse brain anterior and posterior. (A) Annotated"], "order_basis": "paper_text_order:7626", "id": "assay:2"}]}