MichiGAN: sampling from disentangled representations of single-cell data using generative adversarial networks.

MichiGAN: sampling from disentangled representations of single-cell data using generative adversarial networks.
复制标题

DOI:
10.1186/s13059-021-02373-4
复制
发表时间:
2021-05-20
期刊:
影响因子:
12.3
通讯作者:
Welch JD
Welch JD
中科院分区:
生物学1区
文献类型:
--
作者:
Yu H;Welch JD

文献摘要

参考文献

被引文献

相似文献

深度生成模型,如变分自动编码器(VAES)和生成对抗网络(GAN),生成并处理高维图像。我们系统地评估了这些模型在单细胞基因表达数据上的互补优势和劣势。我们还开发了密歇根,这是一种新型的神经网络,它结合了VAE和GANS的优点,在不牺牲数据生成质量的情况下从解开的表示中采样。我们学习了三个大型单细胞RNA-SEQ数据集的解缠表示,并使用密歇根州从这些表示中进行采样。密歇根州允许我们操纵细胞身份的语义不同方面,并预测单细胞基因表达对药物治疗的反应。网上版载有补充材料,可在(10.1186/s13059-021-02373-4)查阅。
Deep generative models such as variational autoencoders (VAEs) and generative adversarial networks (GANs) generate and manipulate high-dimensional images. We systematically assess the complementary strengths and weaknesses of these models on single-cell gene expression data. We also develop MichiGAN, a novel neural network that combines the strengths of VAEs and GANs to sample from disentangled representations without sacrificing data generation quality. We learn disentangled representations of three large single-cell RNA-seq datasets and use MichiGAN to sample from these representations. MichiGAN allows us to manipulate semantically distinct aspects of cellular identity and predict single-cell gene expression response to drug treatment. The online version contains supplementary material available at (10.1186/s13059-021-02373-4).
DOI: 10.1242/dev.173849
发表时间: 2019-06-01
期刊: DEVELOPMENT
影响因子: 4.6
作者:
Bastidas-Ponce, Aimee;Tritschler, Sophie;Bakhti, Mostafa
通讯作者: Bakhti, Mostafa
DOI: 10.1038/s41467-018-04368-5
发表时间: 2018-05-21
影响因子: 16.6
作者:
Ding J;Condon A;Shah SP
通讯作者: Shah SP
DOI: 10.1038/s41587-020-0591-3
发表时间: 2020-08-03
影响因子: 46.9
作者:
Bergen, Volker;Lange, Marius;Theis, Fabian J.
通讯作者: Theis, Fabian J.