BioGAN: Enhancing Transcriptomic Data Generation with Biological Knowledge
Abstract
1. Introduction
- Graph-informed generative modeling for omics data: This study introduces a novel framework that integrates graph neural networks (GNNs) into deep generative models to improve the biological fidelity of synthetic omics data.
- Comprehensive evaluation of synthetic data: The realism and utility of the generated data are assessed through a multi-faceted validation framework, including the following: (a) supervised machine learning models for classification tasks; (b) unsupervised statistical comparisons between real and synthetic data; and (c) feature-level fidelity assessments to quantify biological consistency.
2. Background
2.1. Introduction to Generative Models
2.1.1. Variational Autoencoders
2.1.2. Generative Adversarial Networks
2.2. Foundations of Graphs and Graph Neural Networks
Biological Graphs
2.3. Related Works
3. Materials and Methods
3.1. BioGAN Architecture
3.1.1. Graph-Based Generator
- For a GCN, is undirected, weighted, symmetric, capturing the level of similarity between gene expressions. For each pair of genes and , . The function corr usually corresponds to the Pearson correlation. Several variants may be applied: for example, elements of , whose absolute value is below a certain threshold, are set to zero to avoid spurious correlations. In other cases, is made binary, preserving only pairs of genes displaying strong correlation (or anti-correlation).Example: Suppose genes and show a Pearson correlation of 0.92 across all liver samples. Then, , or 1 if binarized. If gene has a correlation of with , and the threshold is 0.8, then .
- For GRN , is directed and unweighted, capturing which genes regulate the expression of the other. Specifically, if positively regulates , if negatively regulates , and otherwise.Example: If gene is known to activate gene , and repress gene , then , ; notice that regulation is not necessarily reciprocal.
3.1.2. Overall Model Architecture
3.2. Datasets
3.2.1. E. coli Microarray Data
3.2.2. H. sapiens RNA-Seq Data
4. Computational Experiments
4.1. Evaluation Metrics
4.1.1. Unsupervised Metrics
4.1.2. Supervised Metrics—Detectability and Utility
4.1.3. Statistical Tests
4.1.4. Visualization Techniques
4.2. Experiment Design
4.3. Results on E. coli Dataset
4.4. Results on GTEX Dataset
4.4.1. Unsupervised Metrics
4.4.2. Detectability
4.4.3. Utility
4.4.4. Results on Incremental Dataset
- Precision (top left): Both variants of our model maintain consistently high precision as the gene count increases. While WGAN-GP’s control line shows a steady linear increase, its overall performance remains lower than our proposed approaches.
- Recall (top right): All models exhibit a declining trend with an increasing gene count. The GraphConv layer version shows a more gradual descent compared to the WGAN-GP and H2GCN layer versions that drops to approximately 0.3 at 19,075 genes. The H2GCN variant experiences a slightly steeper decline in recall, with a loss of approximately 10% more in performance compared to the WGAN-GP version.
- WD2 (bottom left): All three approaches show increasing Wasserstein scores with higher gene counts, following nearly parallel trajectories. Notably, our model with an H2GCN layer maintains slightly lower WS2 values throughout the range.
- Detectability (bottom right): The GraphConv variant shows increasing detectability as gene count rises, while WGAN-GP’s detectability decreases. The H2GCN layer version stabilizes around 0.80, demonstrating superior performance over WGAN-GP. This suggests our approach generates more realistic data even at higher dimensionalities.
5. Discussions and Conclusions
Future Directions
Author Contributions
Funding
Informed Consent Statement
Data Availability Statement
Acknowledgments
Conflicts of Interest
Appendix A. Dataset Distribution
Appendix B. Visualization
Appendix C. Additional Results on Full Features Spaces
Model | WD2 (↓) | Precision (↑) | Recall (↑) | Correlation (↑) |
---|---|---|---|---|
CVAE | 150.122 ± 0.179 | 0.761 ± 0.005 | 0.002 ± 0.001 | 0.606 ± 0.004 |
GAN | 150.129 ± 0.187 | 0.761 ± 0.006 | 0.002 ± 0.001 | 0.606 ± 0.004 |
WGAN-GP | 102.009 ± 0.461 | 0.935 ± 0.006 | 0.307 ± 0.009 | 0.938 ± 0.002 |
Proposed-GraphConv (0.90) | 103.758 ± 0.296 | 0.987 ± 0.003 | 0.475 ± 0.013 | 0.944 ± 0.002 |
Proposed-GraphConv (0.80) | 109.507 ± 0.251 | 0.942 ± 0.007 | 0.621 ± 0.015 | 0.935 ± 0.001 |
Proposed-H2GCN (0.90) | 98.544 ± 0.314 | 0.997 ± 0.002 | 0.232 ± 0.013 | 0.960 ± 0.001 |
Proposed-H2GCN (0.90) + GRN matrix | 98.962 ± 0.180 | 0.998 ± 0.001 | 0.248 ± 0.019 | 0.959 ± 0.001 |
Entire Feature Space (↓) | PC’s (↓) | |||||
---|---|---|---|---|---|---|
Model | F1 (LR) | F1 (RF) | F1 (MLP) | F1 (LR) | F1 (RF) | F1 (MLP) |
CVAE | 1.000 ± 0.000 | 1.000 ± 0.000 | 1.000 ± 0.000 | 0.990 ± 0.002 | 0.999 ± 0.001 | 0.998 ± 0.001 |
GAN | 1.000 ± 0.000 | 1.000 ± 0.000 | 1.000 ± 0.000 | 0.991 ± 0.001 | 0.999 ± 0.001 | 0.998 ± 0.001 |
WGAN-GP | 0.992 ± 0.001 | 0.988 ± 0.001 | 0.999 ± 0.001 | 0.541 ± 0.018 | 0.971 ± 0.002 | 0.989 ± 0.001 |
Proposed-GraphConv (0.90) | 0.999 ± 0.001 | 1.000 ± 0.000 | 0.991 ± 0.002 | 0.563 ± 0.006 | 0.950 ± 0.003 | 0.981 ± 0.002 |
Proposed-GraphConv (0.80) | 0.999 ± 0.001 | 1.000 ± 0.000 | 0.999 ± 0.0001 | 0.662 ± 0.008 | 0.959 ± 0.005 | 0.978 ± 0.003 |
Proposed-H2GCN (0.90) | 0.798 ± 0.003 | 1.000 ± 0.000 | 0.990 ± 0.003 | 0.524 ± 0.010 | 0.922 ± 0.008 | 0.987 ± 0.003 |
Proposed-H2GCN (0.90) + GRN matrix | 0.831 ± 0.003 | 1.000 ± 0.000 | 0.990 ± 0.001 | 0.528 ± 0.011 | 0.925 ± 0.005 | 0.984 ± 0.004 |
Model | F1 (LR) (↑) | F1 (RF) (↑) | F1 (MLP) (↑) |
---|---|---|---|
CVAE | 0.384 ± 0.02 | 0.435 ± 0.001 | 0.344 ± 0.016 |
GAN | 0.389 ± 0.015 | 0.435 ± 0.001 | 0.345 ± 0.017 |
WGAN-GP | 0.959 ± 0.003 | 0.988 ± 0.003 | 0.940 ± 0.011 |
Proposed-GraphConv (0.90) | 0.975 ± 0.003 | 0.917 ± 0.003 | 0.951 ± 0.006 |
Proposed-GraphConv (0.80) | 0.976 ± 0.001 | 0.928 ± 0.009 | 0.965 ± 0.008 |
Proposed-H2GCN (0.90) | 0.967 ± 0.003 | 0.874 ± 0.006 | 0.949 ± 0.006 |
Proposed-H2GCN (0.90) + GRN matrix | 0.972 ± 0.003 | 0.931 ± 0.007 | 0.946 ± 0.007 |
Appendix D. Additional Figures
References
- Heavey, M.K.; Durmusoglu, D.; Crook, N.; Anselmo, A.C. Discovery and delivery strategies for engineered live biotherapeutic products. Trends Biotechnol. 2022, 40, 354–369. [Google Scholar] [CrossRef]
- Procter, R.; Tolmie, P.; Rouncefield, M. Holding AI to account: Challenges for the delivery of trustworthy AI in healthcare. ACM Trans. Comput.-Hum. Interact. 2023, 30, 1–34. [Google Scholar] [CrossRef]
- Ghaffar Nia, N.; Kaplanoglu, E.; Nasab, A. Evaluation of artificial intelligence techniques in disease diagnosis and prediction. Discov. Artif. Intell. 2023, 3, 5. [Google Scholar] [CrossRef]
- European Union. Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the Protection of Natural Persons with Regard to the Processing of Personal Data and on the Free Movement of Such Data (General Data Protection Regulation); Official Journal of the European Union: Brussels, Belgium, 2016; Available online: https://eur-lex.europa.eu/eli/reg/2016/679/oj (accessed on 8 June 2025).
- Health insurance portability and accountability act of 1996 (HIPAA). Public Law 1996, 104, 191.
- Arjovsky, M.; Chintala, S.; Bottou, L. Wasserstein GAN. arXiv 2017, arXiv:1701.07875v3. [Google Scholar]
- Gulrajani, I.; Ahmed, F.; Arjovsky, M.; Dumoulin, V.; Courville, A.C. Improved Training of Wasserstein GANs. Adv. Neural Inf. Process. Syst. 2017, 30, 5767–5777. [Google Scholar]
- Kingma, D.P.; Welling, M. An introduction to variational autoencoders. Found. Trends® Mach. Learn. 2019, 12, 307–392. [Google Scholar] [CrossRef]
- Mirza, M.; Osindero, S. Conditional generative adversarial nets. arXiv 2014, arXiv:1411.1784. [Google Scholar]
- Stuart, J.M.; Segal, E.; Koller, D.; Kim, S.K. A gene-coexpression network for global discovery of conserved genetic modules. Science 2003, 302, 249–255. [Google Scholar] [CrossRef]
- Karlebach, G.; Shamir, R. Modelling and analysis of gene regulatory networks. Nat. Rev. Mol. Cell Biol. 2008, 9, 770–780. [Google Scholar] [CrossRef]
- Hecker, M.; Lambeck, S.; Toepfer, S.; Van Someren, E.; Guthke, R. Gene regulatory network inference: Data integration in dynamic models—A review. Biosystems 2009, 96, 86–103. [Google Scholar] [CrossRef] [PubMed]
- Viñas, R.; Andrés-Terré, H.; Liò, P.; Bryson, K. Adversarial generation of gene expression data. Bioinformatics 2022, 38, 730–737. [Google Scholar] [CrossRef] [PubMed]
- Van den Bulcke, T.; Van Leemput, K.; Naudts, B.; van Remortel, P.; Ma, H.; Verschoren, A.; De Moor, B.; Marchal, K. SynTReN: A generator of synthetic gene expression data for design and analysis of structure learning algorithms. BMC Bioinform. 2006, 7, 43. [Google Scholar] [CrossRef] [PubMed]
- Schaffter, T.; Marbach, D.; Floreano, D. GeneNetWeaver: In silico benchmark generation and performance profiling of network inference methods. Bioinformatics 2011, 27, 2263–2270. [Google Scholar] [CrossRef] [PubMed]
- Lacan, A.; Sebag, M.; Hanczar, B. GAN-based data augmentation for transcriptomics: Survey and comparative assessment. Bioinformatics 2023, 39, i111–i120. [Google Scholar] [CrossRef]
- Liu, T.; Qian, Z.; Berrevoets, J.; van der Schaar, M. GOGGLE: Generative modelling for tabular data by learning relational structure. In Proceedings of the Eleventh International Conference on Learning Representations, Kigali, Rwanda, 1–5 May 2023. [Google Scholar]
- Zhu, J.; Yan, Y.; Zhao, L.; Heimann, M.; Akoglu, L.; Koutra, D. Beyond homophily in graph neural networks: Current limitations and effective designs. Adv. Neural Inf. Process. Syst. 2020, 33, 7793–7804. [Google Scholar]
- Shao, Y.; Mao, L.; Ye, L.; Li, J.; Yang, P.; Ji, C.; Wu, Z. H2GCN: A hybrid hypergraph convolution network for skeleton-based action recognition. J. King Saud-Univ.-Comput. Inf. Sci. 2024, 36, 102072. [Google Scholar] [CrossRef]
- Morris, C.; Ritzert, M.; Fey, M.; Hamilton, W.L.; Lenssen, J.E.; Rattan, G.; Grohe, M. Weisfeiler and leman go neural: Higher-order graph neural networks. In Proceedings of the AAAI Conference on Artificial Intelligence, Honolulu, HI, USA, 27 January–1 February 2019; Volume 33, pp. 4602–4609. [Google Scholar]
- Team, M. M3D: A Multimodal Medical Dataset. 2025. Available online: http://m3d.mssm.edu/ (accessed on 29 April 2025).
- Lonsdale, J.; Thomas, J.; Salvatore, M.; Phillips, R.; Lo, E.; Shad, S.; Hasz, R.; Walters, G.; Garcia, F.; Young, N.; et al. The genotype-tissue expression (GTEx) project. Nat. Genet. 2013, 45, 580–585. [Google Scholar] [CrossRef]
- González, B.; Gutiérrez-Ríos, R.; Romero, A.; Díaz-Peralta, M.; Colchado, A.; Espinosa, R. RegulonDB: A Database of Transcriptional Regulation in Escherichia coli. 2025. Available online: https://regulondb.ccg.unam.mx/ (accessed on 29 April 2025).
- Kynkäänniemi, T.; Karras, T.; Laine, S.; Lehtinen, J.; Aila, T. Improved precision and recall metric for assessing generative models. In Proceedings of the Advances in Neural Information Processing Systems, Vancouver, BC, Canada, 8–14 December 2019; Curran Associates Inc.: Red Hook, NY, USA, 2019. [Google Scholar]
- McInnes, L.; Healy, J.; Melville, J. Umap: Uniform manifold approximation and projection for dimension reduction. arXiv 2018, arXiv:1802.03426. [Google Scholar]
- Topping, J.; Di Giovanni, F.; Chamberlain, B.P.; Dong, X.; Bronstein, M.M. Understanding over-squashing and bottlenecks on graphs via curvature. arXiv 2021, arXiv:2111.14522. [Google Scholar]
- Vazquez, M.; Krallinger, M.; Leitner, F.; Kuiper, M.; Valencia, A.; Laegreid, A. ExTRI: Extraction of transcription regulation interactions from literature. Biochim. Biophys. Acta-(BBA)-Gene Regul. Mech. 2022, 1865, 194778. [Google Scholar] [CrossRef] [PubMed]
Model | Precision (↑) | Recall (↑) | Correlation (↑) |
---|---|---|---|
CVAE | 0.338 ± 0.038 | 0.041 ± 0.031 | 0.532 ± 0.005 |
GAN | 1.000 ± 0.000 | 0.000 ± 0.000 | 0.019 ± 0.026 |
WGAN-GP | 0.914 ± 0.003 | 0.620 ± 0.016 | 0.809 ± 0.011 |
BioGAN-GraphConv | 0.515 ± 0.024 | 0.917 ± 0.008 | 0.834 ± 0.006 |
BioGAN-H2GCN | 0.487 ± 0.024 | 0.928 ± 0.010 | 0.833 ± 0.008 |
Entire Features Space (↓) | Principal Components (↓) | |||||
---|---|---|---|---|---|---|
Model | F1 (LR) | F1 (RF) | F1 (MLP) | F1 (LR) | F1 (RF) | F1 (MLP) |
CVAE | 0.998 ± 0.001 | 1.000 ± 0.000 | 1.000 ± 0.000 | 0.988 ± 0.002 | 0.992 ± 0.003 | 1.000 ± 0.000 |
GAN | 1.000 ± 0.0000 | 1.000 ± 0.0000 | 1.000 ± 0.0000 | 0.980 ± 0.003 | 0.999 ± 0.001 | 1.000 ± 0.0000 |
WGAN-GP | 1.000 ± 0.0000 | 0.994 ± 0.005 | 1.0000 ± 0.000 | 0.847 ± 0.007 | 0.971 ± 0.004 | 0.995 ± 0.001 |
BioGAN-GraphConv | 0.845 ± 0.026 | 1.000 ± 0.0000 | 0.940 ± 0.015 | 0.571 ± 0.024 | 0.940 ± 0.012 | 0.949 ± 0.007 |
BioGAN-H2GCN | 0.841 ± 0.013 | 1.0000 ± 0.0000 | 0.940 ± 0.010 | 0.569 ± 0.021 | 0.942 ± 0.006 | 0.942 ± 0.011 |
Model | F1 (LR) (↑) | F1 (RF) (↑) | F1 (MLP) (↑) |
---|---|---|---|
CVAE | 0.787 ± 0.059 | 0.647 ± 0.098 | 0.679 ± 0.036 |
GAN | 0.417 ± 0.035 | 0.799 ± 0.000 | 0.296 ± 0.009 |
WGAN-GP | 0.753 ± 0.010 | 0.888 ± 0.002 | 0.776 ± 0.006 |
BioGAN-GraphConv | 0.962 ± 0.005 | 0.877 ± 0.009 | 0.959 ± 0.0091 |
BioGAN-H2GCN | 0.963 ± 0.006 | 0.864 ± 0.008 | 0.958 ± 0.009 |
Model | Precision (↑) | Recall (↑) | Correlation (↑) |
---|---|---|---|
3000 most distinct | |||
CVAE | 0.221 ± 0.017 | 0.181 ± 0.037 | 0.429 ± 0.016 |
GAN | 0.749 ± 0.009 | 0.233 ± 0.032 | 0.752 ± 0.002 |
WGAN-GP | 0.727 ± 0.014 | 0.723 ± 0.010 | 0.936 ± 0.002 |
BioGAN-GraphConv (0.90) | 0.922 ± 0.005 | 0.916 ± 0.006 | 0.973 ± 0.001 |
BioGAN-GraphConv (0.80) | 0.669 ± 0.019 | 0.926 ± 0.006 | 0.958 ± 0.001 |
BioGAN-H2GCN (0.90) | 0.991 ± 0.003 | 0.752 ± 0.015 | 0.981 ± 0.001 |
BioGAN-H2GCN (0.80) | 0.988 ± 0.001 | 0.858 ± 0.009 | 0.983 ± 0.001 |
3000 less distinct | |||
CVAE | 0.379 ± 0.011 | 0.000 ± 0.000 | 0.320 ± 0.005 |
GAN | 0.939 ± 0.003 | 0.000 ± 0.000 | 0.322 ± 0.001 |
WGAN-GP | 0.931 ± 0.005 | 0.012 ± 0.003 | 0.632 ± 0.002 |
BioGAN-GraphConv (0.90) | 0.928 ± 0.007 | 0.066 ± 0.008 | 0.705 ± 0.003 |
BioGAN-GraphConv (0.80) | 0.866 ± 0.011 | 0.124 ± 0.012 | 0.716 ± 0.003 |
BioGAN-H2GCN (0.90) | 0.998 ± 0.001 | 0.005 ± 0.000 | 0.749 ± 0.001 |
BioGAN-H2GCN (0.80) | 0.997 ± 0.001 | 0.002 ± 0.001 | 0.738 ± 0.001 |
All genes | |||
CVAE | 0.761 ± 0.005 | 0.002 ± 0.001 | 0.606 ± 0.004 |
GAN | 0.761 ± 0.006 | 0.002 ± 0.001 | 0.606 ± 0.004 |
WGAN-GP | 0.935 ± 0.006 | 0.307 ± 0.009 | 0.938 ± 0.002 |
BioGAN-GraphConv (0.90) | 0.987 ± 0.003 | 0.475 ± 0.013 | 0.944 ± 0.002 |
BioGAN-GraphConv (0.80) | 0.942 ± 0.007 | 0.621 ± 0.015 | 0.935 ± 0.001 |
BioGAN-H2GCN (0.90) | 0.997 ± 0.002 | 0.232 ± 0.013 | 0.960 ± 0.001 |
BioGAN-H2GCN (0.80) | 0.997 ± 0.001 | 0.002 ± 0.001 | 0.738 ± 0.001 |
Entire Features Space ↓ | PC’s ↓ | |||||
---|---|---|---|---|---|---|
Model | F1 (LR) | F1 (RF) | F1 (MLP) | F1 (LR) | F1 (RF) | F1 (MLP) |
3000 most distinct | ||||||
CVAE | 0.995 ± 0.001 | 1.000 ± 0.001 | 0.999 ± 0.001 | 0.921 ± 0.005 | 0.998 ± 0.001 | 0.998 ± 0.001 |
GAN | 1.00 ± 0.00 | 0.997 ± 0.00 | 1.00 ± 0.00 | 0.939 ± 0.004 | 0.992 ± 0.003 | 0.998 ± 0.000 |
WGAN-GP | 0.992 ± 0.001 | 0.988 ± 0.001 | 0.999 ± 0.001 | 0.556 ± 0.012 | 0.972 ± 0.003 | 0.996 ± 0.0017 |
BioGAN-GraphConv (0.90) | 0.978 ± 0.003 | 0.999 ± 0.001 | 0.991 ± 0.002 | 0.567 ± 0.011 | 0.936 ± 0.006 | 0.991 ± 0.002 |
BioGAN-GraphConv (0.80) | 0.999 ± 0.001 | 0.999 ± 0.001 | 0.999 ± 0.001 | 0.671 ± 0.007 | 0.896 ± 0.006 | 0.985 ± 0.002 |
BioGAN-H2GCN (0.90) | 0.766 ± 0.005 | 0.934 ± 0.006 | 0.996 ± 0.001 | 0.522 ± 0.009 | 0.919 ± 0.006 | 0.995 ± 0.002 |
BioGAN-H2GCN (0.80) | 0.875 ± 0.003 | 0.997 ± 0.001 | 0.994 ± 0.002 | 0.5359 ± 0.004 | 0.920 ± 0.009 | 0.994 ± 0.002 |
All genes | ||||||
CVAE | 1.000 ± 0.000 | 1.000 ± 0.000 | 1.000 ± 0.000 | 0.990 ± 0.002 | 0.999 ± 0.001 | 0.998 ± 0.001 |
GAN | 1.000 ± 0.000 | 1.000 ± 0.000 | 1.000 ± 0.000 | 0.991 ± 0.001 | 0.999 ± 0.001 | 0.998 ± 0.001 |
WGAN-GP | 0.992 ± 0.001 | 0.988 ± 0.001 | 0.999 ± 0.001 | 0.541 ± 0.018 | 0.971 ± 0.002 | 0.989 ± 0.001 |
BioGAN-GraphConv (0.90) | 0.999 ± 0.001 | 1.000 ± 0.000 | 0.991 ± 0.002 | 0.563 ± 0.006 | 0.950 ± 0.003 | 0.981 ± 0.002 |
BioGAN-GraphConv (0.80) | 0.999 ± 0.001 | 1.000 ± 0.000 | 0.999 ± 0.0001 | 0.662 ± 0.008 | 0.959 ± 0.005 | 0.978 ± 0.003 |
BioGAN-H2GCN (0.90) | 0.798 ± 0.003 | 1.000 ± 0.000 | 0.990 ± 0.003 | 0.524 ± 0.010 | 0.922 ± 0.008 | 0.987 ± 0.003 |
Model | F1 (LR) (↑) | F1 (RF) (↑) | F1 (MLP) (↑) |
---|---|---|---|
3000 most distinct | |||
CVAE | 0.921 ± 0.012 | 0.186 ± 0.008 | 0.705 ± 0.018 |
GAN | 0.809 ± 0.001 | 0.842 ± 0.007 | 0.797 ± 0.010 |
WGAN-GP | 0.925 ± 0.001 | 0.940 ± 0.003 | 0.799 ± 0.04 |
Proposed-GraphConv (0.90) | 0.968 ± 0.002 | 0.939 ± 0.005 | 0.955 ± 0.0023 |
Proposed-GraphConv (0.80) | 0.975 ± 0.002 | 0.957 ± 0.004 | 0.967 ± 0.003 |
Proposed-H2GCN (0.90) | 0.958 ± 0.003 | 0.929 ± 0.004 | 0.9505 ± 0.002 |
Proposed-H2GCN (0.80) | 0.966 ± 0.003 | 0.934 ± 0.003 | 0.960 ± 0.001 |
All genes | |||
CVAE | 0.384 ± 0.02 | 0.435 ± 0.001 | 0.344 ± 0.016 |
GAN | 0.389 ± 0.015 | 0.435 ± 0.001 | 0.345 ± 0.017 |
WGAN-GP | 0.959 ± 0.003 | 0.988 ± 0.003 | 0.940 ± 0.011 |
BioGAN-GraphConv (0.90) | 0.975 ± 0.003 | 0.917 ± 0.003 | 0.951 ± 0.006 |
BioGAN-GraphConv (0.80) | 0.976 ± 0.001 | 0.928 ± 0.009 | 0.965 ± 0.008 |
BioGAN-H2GCN (0.90) | 0.967 ± 0.003 | 0.874 ± 0.006 | 0.949 ± 0.006 |
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
© 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).
Share and Cite
Panaccione, F.P.; Mongardi, S.; Masseroli, M.; Pinoli, P. BioGAN: Enhancing Transcriptomic Data Generation with Biological Knowledge. Bioengineering 2025, 12, 658. https://doi.org/10.3390/bioengineering12060658
Panaccione FP, Mongardi S, Masseroli M, Pinoli P. BioGAN: Enhancing Transcriptomic Data Generation with Biological Knowledge. Bioengineering. 2025; 12(6):658. https://doi.org/10.3390/bioengineering12060658
Chicago/Turabian StylePanaccione, Francesca Pia, Sofia Mongardi, Marco Masseroli, and Pietro Pinoli. 2025. "BioGAN: Enhancing Transcriptomic Data Generation with Biological Knowledge" Bioengineering 12, no. 6: 658. https://doi.org/10.3390/bioengineering12060658
APA StylePanaccione, F. P., Mongardi, S., Masseroli, M., & Pinoli, P. (2025). BioGAN: Enhancing Transcriptomic Data Generation with Biological Knowledge. Bioengineering, 12(6), 658. https://doi.org/10.3390/bioengineering12060658