用于精准预测空间蛋白组与转录组细胞丰度的智能方法

doi:10.3778/j.issn.1002-8331.2409-0115

摘要/Abstract

摘要： 细胞类型反卷积是从复杂组学数据（如空间蛋白组学和转录组学）中解析细胞丰度的关键技术。然而，传统的反卷积方法通常仅针对特定组学数据进行设计，未能提供通用的预测框架。为此，提出了一种新型深度学习反卷积方法——SpatialDec，旨在同时处理蛋白组学和转录组学数据。SpatialDec结合了图神经网络的优势，能够识别复杂的细胞交互关系，并通过三元组损失函数强化细胞间的相似性与差异性。此外，SpatialDec采用领域自适应技术，有效去除源数据和目标数据之间的批次效应。大量实验证明，SpatialDec在多个蛋白组学和转录组学数据集上显著提升了预测性能。在蛋白组学数据集上，SpatialDec将均方根误差（RMSE）降低了21.9%，Lin的一致性相关系数（CCC）提高了46.1%，皮尔逊相关系数（PCC）提高了18.0%。在转录组学数据集上，SpatialDec也显著提高了性能，将RMSE降低了34.0%，CCC提高了45.4%，PCC提高了43.8%。

关键词: 细胞类型反卷积, 蛋白组学, 图神经网络, 三元组损失

Abstract: Cell type deconvolution is a pivotal analytical approach that deciphers the abundance of cell types from complex omics datasets, including spatial proteomics and transcriptomics. Traditionally, deconvolution techniques have been tailored to individual omics types, addressing specific research questions but falling short of offering a unified predictive framework. This paper introduces SpatialDec, a deep learning-based deconvolution method applicable to both proteomics and transcriptomics data. SpatialDec harnesses the power of graph neural networks to discern intricate cellular interactions and enhances the similarities and differences of cells through triplet loss. Additionally, the domain adaptation technique is employed to remove batch effects between source and target data. Extensive experiments demonstrate that SpatialDec exhibits superior performance across multiple proteomics and transcriptomics datasets. Compared to state-of-the-art methods, SpatialDec reduces root mean square error (RMSE) by 21.9%, increases Lin’s concordance correlation coefficient (CCC) by 46.1%, and boosts Pearson correlation coefficient (PCC) by 18.0% in proteomics datasets. Moreover, SpatialDec decreases RMSE by 34.0%, increases CCC by 45.4%, and improves PCC by 43.8% in transcriptomics datasets.

Key words: cell type deconvolution, proteomics, graph neural networks, triplet loss

宋渝杰, 曾远松, 杨跃东, 戴智明. 用于精准预测空间蛋白组与转录组细胞丰度的智能方法[J]. 计算机工程与应用, 2025, 61(13): 176-184.

SONG Yujie, ZENG Yuansong, YANG Yuedong, DAI Zhiming. Smart Method for Precise Prediction of Cell Abundance in Spatial Proteomics and Transcriptomics[J]. Computer Engineering and Applications, 2025, 61(13): 176-184.

参考文献

[1] JIANG S, QIAN Q H, ZHU T T, et al. Cell Taxonomy: a curated repository of cell types with multifaceted characterization[J]. Nucleic Acids Research, 2023, 51(D1): D853-D860.
[2] ZEISEL A, HOCHGERNER H, L?NNERBERG P, et al. Molecular architecture of the mouse nervous system[J]. Cell, 2018, 174(4): 999-1014.
[3] MOFFITT J R, BAMBAH-MUKKU D, EICHHORN S W, et al. Molecular, spatial, and functional single-cell profiling of the hypothalamic preoptic region[J]. Science, 2018, 362(6416): eaau5324.
[4] RAO A, BARKLEY D, FRAN?A G S, et al. Exploring tissue architecture using spatial transcriptomics[J]. Nature, 2021, 596(7871): 211-220.
[5] MOSES L, PACHTER L. Museum of spatial transcriptomics[J]. Nature Methods, 2022, 19(5): 534-546.
[6] HAN X P, WANG R Y, ZHOU Y C, et al. Mapping the mouse cell atlas by microwell-seq[J]. Cell, 2018, 172(5): 1091-1107.
[7] CAO J Y, SPIELMANN M, QIU X J, et al. The single-cell transcriptional landscape of mammalian organogenesis[J]. Nature, 2019, 566(7745): 496-502.
[8] SVENSSON V, VENTO-TORMO R, TEICHMANN S A. Exponential scaling of single-cell RNA-seq in the past decade[J]. Nature Protocols, 2018, 13(4): 599-604.
[9] NEWMAN A M, STEEN C B, LIU C L, et al. Determining cell type abundance and expression from bulk tissues with digital cytometry[J]. Nature Biotechnology, 2019, 37(7): 773-782.
[10] MENDEN K, MAROUF M, OLLER S, et al. Deep learning-based cell composition analysis from tissue expression profiles[J]. Science Advances, 2020, 6(30): eaba2619.
[11] CABLE D M, MURRAY E, ZOU L S, et al. Robust decomposition of cell type mixtures in spatial transcriptomics[J]. Nature Biotechnology, 2022, 40(4): 517-526.
[12] EDWARDS N J, OBERTI M, THANGUDU R R, et al. The CPTAC data portal: a resource for cancer proteomics research[J]. Journal of Proteome Research, 2015, 14(6): 2707-2713.
[13] LOPEZ R, LI B G, KEREN-SHAUL H, et al. DestVI identifies continuums of cell types in spatial transcriptomics data[J]. Nature Biotechnology, 2022, 40(9): 1360-1369.
[14] CHU T, WANG Z, PE’ER D, et al. Cell type and gene expression deconvolution with BayesPrism enables Bayesian integrative analysis across bulk and single-cell RNA sequencing in oncology[J]. Nature Cancer, 2022, 3(4): 505-517.
[15] WANG X R, PARK J, SUSZTAK K, et al. Bulk tissue cell type deconvolution with multi-subject single-cell expression reference[J]. Nature Communications, 2019, 10(1): 380.
[16] BIANCALANI T, SCALIA G, BUFFONI L, et al. Deep learning and alignment of spatially resolved single-cell transcriptomes with Tangram[J]. Nature Methods, 2021, 18(11): 1352-1362.
[17] WANG F, YANG F, HUANG L K, et al. Deep domain adversarial neural network for the deconvolution of cell type mixtures in tissue proteome profiling[J]. Nature Machine Intelligence, 2023, 5(11): 1236-1249.
[18] HINTON G, DENG L, YU D, et al. Deep neural networks for acoustic modeling in speech recognition: the shared views of four research groups[J]. IEEE Signal Processing Magazine, 2012, 29(6): 82-97.
[19] MNIH V. Playing atari with deep reinforcement learning[J]. arXiv:1312.5602, 2013.
[20] KRIZHEVSKY A, SUTSKEVER I, HINTON G E. ImageNet classification with deep convolutional neural networks[C]//Advances in Neural Information Processing Systems, 2012.
[21] RAO J H, ZHOU X, LU Y T, et al. Imputing single-cell RNA-seq data by combining graph convolution and autoencoder neural networks[J]. iScience, 2021, 24(5): 102393.
[22] ROOHANI Y, HUANG K X, LESKOVEC J. Predicting transcriptional outcomes of novel multigene perturbations with GEARS[J]. Nature Biotechnology, 2024, 42(6): 927-935.
[23] PAN Y, WANG X, SUN J, et al. Multimodal joint deconvolution and integrative signature selection in protomics[J]. Communications Biology, 2024, 7(1): 493.
[24] TENG P N, SCHAAF J P, ABULEZ T, et al. ProteoMixture: a cell type deconvolution tool for bulk tissue proteomic data[J]. iScience, 2024, 27(3): 109198.
[25] XU B, WANG N Y, CHEN T Q, et al. Empirical evaluation of rectified activations in convolutional network[J]. arXiv: 1505.00853, 2015.
[26] HAMILTON W L, YING R, LESKOVEC J. Inductive representation learning on large graphs[C]//Advances in Neural Information Processing Systems, 2017.
[27] SCHROFF F, KALENICHENKO D, PHILBIN J. FaceNet: a unified embedding for face recognition and clustering[C]//Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 815-823.
[28] DOU M W, CLAIR G, TSAI C F, et al. High-throughput single cell proteomics enabled by multiplex isobaric labeling in a nanodroplet sample preparation platform[J]. Analytical Chemistry, 2019, 91(20): 13119-13127.
[29] WOO J, WILLIAMS S M, MARKILLIE L M, et al. High-throughput and high-efficiency sample preparation for single-cell proteomics using a nested nanowell chip[J]. Nature Communications, 2021, 12(1): 6246.
[30] GRAY G K, LI C M, ROSENBLUTH J M, et al. A human breast atlas integrating single-cell proteomics and transcriptomics[J]. Developmental Cell, 2022, 57(11): 1400-1420.
[31] LIN L I. A concordance correlation coefficient to evaluate reproducibility[J]. Biometrics, 1989, 45(1): 255-268.
[32] KARL P. Contributions to the mathematical theory of evolution[J]. Philosophical Transactions of the Royal Society of London Series A, 1894, 185: 71-110.