{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T18:50:32Z","timestamp":1791226232329,"version":"4.1.0"},"reference-count":39,"publisher":"Public Library of Science (PLoS)","issue":"10","license":[{"start":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T00:00:00Z","timestamp":1791158400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["www.ploscompbiol.org"],"crossmark-restriction":false},"short-container-title":["PLoS Comput Biol"],"abstract":"<jats:p>\n                    A central challenge in downstream single-cell RNA sequencing (scRNA-seq) analysis is the high-dimensional, small-sample (HDSS) regime, often compounded by class imbalance from rare cell types. These factors hinder robust feature (gene) selection and cell clustering and limit the realism of samples generated by existing simulators. We introduce\n                    <jats:italic>GARAGE<\/jats:italic>\n                    , a\n                    <jats:bold>G<\/jats:bold>\n                    raph-\n                    <jats:bold>A<\/jats:bold>\n                    ttentive\n                    <jats:bold>RA<\/jats:bold>\n                    re-cell aware single-cell data\n                    <jats:bold>GE<\/jats:bold>\n                    neration that augments the generator\u2019s input with a small, attention-weighted\n                    <jats:italic>\u2018leakage\u2019<\/jats:italic>\n                    of real cells in addition to prior noise. Specifically, we build a\n                    <jats:italic>k<\/jats:italic>\n                    -nearest-neighbour cell graph and use a graph attention network (GAT) to prioritize nodes that likely represent under-sampled (rare) subpopulations; these high-attention cell embeddings are injected into the generator input to steer synthesis toward biologically plausible regions of the data manifold while respecting cell-type proportions. This attention-guided leakage accelerates training, reduces mode dropping, and yields realistic synthetic cells that preserve rare-cell structure. Across real scRNA-seq benchmarks, GARAGE improves downstream feature selection and clustering compared with state-of-the-art baselines. In summary, GARAGE directly addresses HDSS and rarity in scRNA-seq by coupling graph attention with adversarial generation to produce high-fidelity synthetic cells that enhance downstream analyses. The corresponding software is available at:\n                    <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/RitwikGanguly\/GARAGE\" xlink:type=\"simple\">https:\/\/github.com\/RitwikGanguly\/GARAGE<\/jats:ext-link>\n                    .\n                  <\/jats:p>","DOI":"10.1371\/journal.pcbi.1014601","type":"journal-article","created":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T17:51:10Z","timestamp":1791222670000},"page":"e1014601","update-policy":"https:\/\/doi.org\/10.1371\/journal.pcbi.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["A graph-attentive GAN for rare-cell-aware single-cell RNA-seq data generation"],"prefix":"10.1371","volume":"22","author":[{"ORCID":"https:\/\/orcid.org\/0009-0000-4626-3760","authenticated-orcid":true,"given":"Ritwik","family":"Ganguly","sequence":"first","affiliation":[{"name":"Aliah University","place":["Kolkata, India"],"department":["Computer Science and Engineering"]},{"name":"Indraprastha Institute of Information Technology","place":["New Delhi, India"],"department":["Computational Biology"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0000-3520-9216","authenticated-orcid":true,"given":"Sana","family":"Aafrine","sequence":"additional","affiliation":[{"name":"Aliah University","place":["Kolkata, India"],"department":["Computer Science and Engineering"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0003-0174-2359","authenticated-orcid":true,"given":"Sk Md Mosaddek","family":"Hossain","sequence":"additional","affiliation":[{"name":"Aliah University","place":["Kolkata, India"],"department":["Computer Science and Engineering"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-3371-8516","authenticated-orcid":true,"given":"Sumanta","family":"Ray","sequence":"additional","affiliation":[{"name":"The West Bengal National University of Juridical Sciences","place":["Kolkata, India"],"department":["Data Science"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,10,5]]},"reference":[{"issue":"6","key":"pcbi.1014601.ref001","doi-asserted-by":"crossref","first-page":"565","DOI":"10.1038\/nmeth.4292","article-title":"Normalizing single-cell RNA sequencing data: Challenges and opportunities","volume":"14","author":"CA Vallejos","year":"2017","journal-title":"Nat Methods"},{"key":"pcbi.1014601.ref002","article-title":"Markercapsule: Explainable single cell typing using capsule networks","author":"S Ray","year":"2020","journal-title":"bioRxiv"},{"key":"pcbi.1014601.ref003","doi-asserted-by":"crossref","DOI":"10.7554\/eLife.27041","article-title":"The human cell atlas","volume":"6","author":"A Regev","year":"2017","journal-title":"Elife"},{"issue":"1","key":"pcbi.1014601.ref004","doi-asserted-by":"crossref","first-page":"47","DOI":"10.1186\/s13059-018-1426-0","article-title":"Mapping human pluripotent stem cell differentiation pathways using high throughput single-cell RNA-sequencing","volume":"19","author":"X Han","year":"2018","journal-title":"Genome Biol"},{"issue":"5","key":"pcbi.1014601.ref005","doi-asserted-by":"crossref","first-page":"365","DOI":"10.1038\/nrn3475","article-title":"Power failure: Why small sample size undermines the reliability of neuroscience","volume":"14","author":"KS Button","year":"2013","journal-title":"Nat Rev Neurosci"},{"issue":"5","key":"pcbi.1014601.ref006","doi-asserted-by":"crossref","first-page":"495","DOI":"10.1038\/nbt.3192","article-title":"Spatial reconstruction of single-cell gene expression data","volume":"33","author":"R Satija","year":"2015","journal-title":"Nat Biotechnol"},{"issue":"5","key":"pcbi.1014601.ref007","doi-asserted-by":"crossref","first-page":"273","DOI":"10.1038\/s41576-018-0088-9","article-title":"Challenges in unsupervised clustering of single-cell RNA-seq data","volume":"20","author":"VY Kiselev","year":"2019","journal-title":"Nat Rev Genet"},{"issue":"1","key":"pcbi.1014601.ref008","doi-asserted-by":"crossref","first-page":"321","DOI":"10.1186\/s13059-021-02544-3","article-title":"Feature selection revisited in the single-cell era","volume":"22","author":"P Yang","year":"2021","journal-title":"Genome Biol"},{"issue":"5","key":"pcbi.1014601.ref009","article-title":"Accurate feature selection improves single-cell RNA-seq cell clustering","volume":"22","author":"K Su","year":"2021","journal-title":"Brief Bioinform"},{"key":"pcbi.1014601.ref010","first-page":"2973","volume-title":"Worst-Case Discriminative Feature Selection","author":"S Liao","year":"2019"},{"issue":"3","key":"pcbi.1014601.ref011","doi-asserted-by":"crossref","first-page":"246","DOI":"10.1049\/nbt2.12115","article-title":"scRNA-seq data analysis method to improve analysis performance","volume":"17","author":"J Lu","year":"2023","journal-title":"IET Nanobiotechnol"},{"key":"pcbi.1014601.ref012","article-title":"Generative adversarial nets","volume":"27","author":"I Goodfellow","year":"2014","journal-title":"Adv Neural Inform Process Syst"},{"key":"pcbi.1014601.ref013","first-page":"214","article-title":"Wasserstein generative adversarial networks.","volume-title":"International conference on machine learning","author":"M Arjovsky","year":"2017"},{"key":"pcbi.1014601.ref014","article-title":"f-gan: Training generative neural samplers using variational divergence minimization","volume":"29","author":"S Nowozin","year":"2016","journal-title":"Adv Neural Inform Process Syst"},{"key":"pcbi.1014601.ref015","article-title":"Auto-encoding variational bayes","author":"DP Kingma","year":"2013","journal-title":"arXiv preprint"},{"issue":"1","key":"pcbi.1014601.ref016","doi-asserted-by":"crossref","first-page":"174","DOI":"10.1186\/s13059-017-1305-0","article-title":"Splatter: simulation of single-cell RNA sequencing data","volume":"18","author":"L Zappia","year":"2017","journal-title":"Genome Biol"},{"issue":"1","key":"pcbi.1014601.ref017","doi-asserted-by":"crossref","first-page":"577","DOI":"10.1038\/s42003-022-03473-y","article-title":"LSH-GAN enables in-silico generation of cells for small sample high dimensional scRNA-seq data","volume":"5","author":"S Lall","year":"2022","journal-title":"Commun Biol"},{"key":"pcbi.1014601.ref018","doi-asserted-by":"crossref","unstructured":"Dong W, Moses C, Li K. Efficient k-nearest neighbor graph construction for generic similarity measures. In: Proceedings of the 20th international conference on World wide web; 2011. pp. 577\u201386.","DOI":"10.1145\/1963405.1963487"},{"key":"pcbi.1014601.ref019","unstructured":"Veli\u010dkovi\u0107 P, Cucurull G, Casanova A, Romero A, Lio P, Bengio Y. Graph attention networks. 2017. https:\/\/arxiv.org\/abs\/1710.10903"},{"issue":"1","key":"pcbi.1014601.ref020","doi-asserted-by":"crossref","first-page":"61","DOI":"10.1109\/TNN.2008.2005605","article-title":"The graph neural network model","volume":"20","author":"F Scarselli","year":"2009","journal-title":"IEEE Trans Neural Netw"},{"key":"pcbi.1014601.ref021","doi-asserted-by":"crossref","first-page":"99","DOI":"10.1016\/S0092-8240(05)80006-0","article-title":"A logical calculus of the ideas immanent in nervous activity. 1943","volume":"52","author":"WS McCulloch","year":"1990","journal-title":"Bull Math Biol"},{"issue":"3","key":"pcbi.1014601.ref022","doi-asserted-by":"crossref","first-page":"331","DOI":"10.1007\/s41095-022-0271-y","article-title":"Attention mechanisms in computer vision: A survey","volume":"8","author":"M-H Guo","year":"2022","journal-title":"Comp Visual Med"},{"key":"pcbi.1014601.ref023","article-title":"Semi-supervised classification with graph convolutional networks","author":"TN Kipf","year":"2016","journal-title":"arXiv preprint"},{"issue":"10","key":"pcbi.1014601.ref024","doi-asserted-by":"crossref","first-page":"428","DOI":"10.1016\/j.tics.2007.09.004","article-title":"Learning multiple layers of representation","volume":"11","author":"GE Hinton","year":"2007","journal-title":"Trends Cogn Sci"},{"key":"pcbi.1014601.ref025","doi-asserted-by":"crossref","first-page":"100285","DOI":"10.1016\/j.cosrev.2020.100285","article-title":"A comprehensive survey and analysis of generative models in machine learning","volume":"38","author":"H GM","year":"2020","journal-title":"Comput Sci Rev"},{"key":"pcbi.1014601.ref026","doi-asserted-by":"crossref","unstructured":"Awad M, Khanna R, Awad M, Khanna R. Hidden markov model. Efficient Learning Machines: Theories, Concepts, and Applications for Engineers and System Designers. 2015. pp. 81\u2013104.","DOI":"10.1007\/978-1-4302-5990-9_5"},{"key":"pcbi.1014601.ref027","unstructured":"Viroli C, McLachlan GJ. Deep Gaussian Mixture Models. 2017. Available from: https:\/\/arxiv.org\/abs\/1711.06929"},{"issue":"2","key":"pcbi.1014601.ref028","doi-asserted-by":"crossref","first-page":"307","DOI":"10.1016\/S0016-0032(96)00063-4","article-title":"The Jensen-Shannon divergence","volume":"334","author":"ML Men\u00e9ndez","year":"1997","journal-title":"J Franklin Inst"},{"key":"pcbi.1014601.ref029","doi-asserted-by":"crossref","first-page":"14","DOI":"10.1007\/978-3-642-33275-3_2","article-title":"An introduction to restricted Boltzmann machines.","volume-title":"Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications: 17th Iberoamerican Congress, CIARP 2012, Buenos Aires, Argentina, September 3-6, 2012. Proceedings 17","author":"A Fischer","year":"2012"},{"issue":"7","key":"pcbi.1014601.ref030","doi-asserted-by":"crossref","first-page":"1527","DOI":"10.1162\/neco.2006.18.7.1527","article-title":"A fast learning algorithm for deep belief nets","volume":"18","author":"GE Hinton","year":"2006","journal-title":"Neural Comput"},{"key":"pcbi.1014601.ref031","doi-asserted-by":"crossref","unstructured":"Vincent P, Larochelle H, Bengio Y, Manzagol PA. Extracting and composing robust features with denoising autoencoders. In: Proceedings of the 25th international conference on Machine learning. 2008. pp. 1096\u2013103.","DOI":"10.1145\/1390156.1390294"},{"issue":"9","key":"pcbi.1014601.ref032","doi-asserted-by":"crossref","first-page":"1131","DOI":"10.1038\/nsmb.2660","article-title":"Single-cell RNA-Seq profiling of human preimplantation embryos and embryonic stem cells","volume":"20","author":"L Yan","year":"2013","journal-title":"Nat Struct Mol Biol"},{"issue":"9","key":"pcbi.1014601.ref033","doi-asserted-by":"crossref","first-page":"865","DOI":"10.1038\/nmeth.4380","article-title":"Simultaneous epitope and transcriptome measurement in single cells","volume":"14","author":"M Stoeckius","year":"2017","journal-title":"Nat Methods"},{"issue":"10","key":"pcbi.1014601.ref034","doi-asserted-by":"crossref","first-page":"1053","DOI":"10.1038\/nbt.2967","article-title":"Low-coverage single-cell mRNA sequencing reveals cellular heterogeneity and activated signaling pathways in developing cerebral cortex","volume":"32","author":"AA Pollen","year":"2014","journal-title":"Nat Biotechnol"},{"issue":"4","key":"pcbi.1014601.ref035","article-title":"A single-cell transcriptome atlas of the human pancreas","volume":"3","author":"MJ Muraro","year":"2016","journal-title":"Cell Syst"},{"issue":"8","key":"pcbi.1014601.ref036","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1038\/s12276-018-0071-8","article-title":"Single-cell RNA sequencing technologies and bioinformatics pipelines","volume":"50","author":"B Hwang","year":"2018","journal-title":"Exp Mol Med"},{"issue":"6","key":"pcbi.1014601.ref037","doi-asserted-by":"crossref","first-page":"479","DOI":"10.1038\/s41592-019-0425-8","article-title":"Benchmarking single cell RNA-sequencing analysis pipelines using mixture control experiments","volume":"16","author":"L Tian","year":"2019","journal-title":"Nat Methods"},{"issue":"1","key":"pcbi.1014601.ref038","doi-asserted-by":"crossref","first-page":"15","DOI":"10.1186\/s13059-017-1382-0","article-title":"SCANPY: Large-scale single-cell gene expression data analysis","volume":"19","author":"FA Wolf","year":"2018","journal-title":"Genome Biol"},{"issue":"1","key":"pcbi.1014601.ref039","doi-asserted-by":"crossref","first-page":"117","DOI":"10.1007\/BF00532240","article-title":"The Wasserstein distance and approximation theorems","volume":"70","author":"L R\u00fcschendorf","year":"1985","journal-title":"Probab Theory Relat Fields"}],"container-title":["PLOS Computational Biology"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1014601","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T17:51:17Z","timestamp":1791222677000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pcbi.1014601"}},"subtitle":[],"editor":[{"given":"Zhixiang","family":"Lin","sequence":"first","affiliation":[{"name":"The Chinese University of Hong Kong Faculty of Science","place":["HONG KONG"]}],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,10,5]]},"references-count":39,"journal-issue":{"issue":"10","published-online":{"date-parts":[[2026,10,5]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pcbi.1014601","relation":{},"ISSN":["1553-7358"],"issn-type":[{"value":"1553-7358","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,10,5]]}}}