BayReL: Bayesian Relational Learning for Multi-omics Data Integration

Ehsan Hajiramezanali
Ehsan Hajiramezanali
Arman Hasanzadeh
Arman Hasanzadeh

NIPS 2020, 2020.

Cited by: 0|Views28
EI
Weibo:
We have proposed Bayesian Relational Learning, a novel Bayesian relational representation learning method that infers interactions across multi-omics data types

Abstract:

High-throughput molecular profiling technologies have produced high-dimensional multi-omics data, enabling systematic understanding of living systems at the genome scale. Studying molecular interactions across different data types helps reveal signal transduction mechanisms across different classes of molecules. In this paper, we develo...More

Code:

Data:

0
Full Text
Bibtex
Weibo
Introduction
  • Modern high-throughput molecular profiling technologies have produced rich high-dimensional data for different bio-molecules at the genome, constituting genome, transcriptome, translatome, proteome, metabolome, epigenome, and interactome scales [Huang et al, 2017, Hajiramezanali et al, 2018b, 2019b, Karimi et al, 2020, Pakbin et al, 2018]
  • Such multi-view data span a diverse range of cellular activities, developing an understanding of how these data types quantitatively relate to each other and to phenotypic characteristics remains elusive.
  • The presented work contains three major contributions: 1) The authors propose a novel Bayesian relation learning framework, BayReL, that can flexibly incorporate the available graph dependency structure of each view. 2) It can exploit non-linear transformations and provide probabilistic interpretation simultaneously. 3) It can infer interactions across different heterogeneous features of input datasets, which is critical to derive meaningful biological knowledge for integrative multi-omics data analysis
Highlights
  • Modern high-throughput molecular profiling technologies have produced rich high-dimensional data for different bio-molecules at the genome, constituting genome, transcriptome, translatome, proteome, metabolome, epigenome, and interactome scales [Huang et al, 2017, Hajiramezanali et al, 2018b, 2019b, Karimi et al, 2020, Pakbin et al, 2018]
  • We evaluate Bayesian Relational Learning (BayReL) and baselines in two metrics – 1) accuracy to identify the validated molecules interacting with P. aeruginosa which will be referred as positive accuracy, 2) accuracy of not detecting common targets between anaerobic microbes and notable pathogen which we refer to this measure as negative accuracy
  • Using 50% of all the samples, while the average prediction sensitivity of BayReL reduces less than 2% in the worst case scenario, Bayesian CCA (BCCA)’s performance degraded around 6%
  • The results prove that BayReL performs better than BCCA with fewer number of observed samples
  • We have proposed BayReL, a novel Bayesian relational representation learning method that infers interactions across multi-omics data types
  • BayReL is a generative model with Bayesian modeling and robust variational inference and is equipped with natural uncertainty estimates, which will help derive reproducible and accurate prediction for robust decision making, with the ultimate goal of improving human health outcomes, as showcased in the three experiments
Methods
  • The authors propose a new graph-structured data integration method, Bayesian Relational Learning (BayReL), for integrative analysis of multi-omics data.
  • The goal of the model is to find inter-relations between nodes of the graphs in different views.
  • The authors model these relations as edges of a multi-partite graph G.
  • The nodes in the multi-partite graph G are the union of the nodes in all views, i.e. VG =
Results
  • Considering higher than 97% negative accuracy, the best positive accuracy of BayReL, BCCA, and SRCA are 82.7% ± 4.7, 28.30% ± 3.21, and 26.41%, respectively.
  • BayReL substantially outperforms the baselines with up to 54% margin.
  • Using 50% of all the samples, while the average prediction sensitivity of BayReL reduces less than 2% in the worst case scenario, BCCA’s performance degraded around 6%.
  • Following Lee et al [2018], the authors study 53 out of 160 drugs that have less than 50% cell viability in at least half of the patient samples
Conclusion
  • The authors have proposed BayReL, a novel Bayesian relational representation learning method that infers interactions across multi-omics data types.
  • BayReL is unique in its model and potential applications
  • This novel generative model is able to deal with growing complexity and heterogeneity of modern large-scale data with complex dependency structures, which is especially critical when analyzing multi-omics data to derive biological insights, the main focus of the research.
  • BayReL is a generative model with Bayesian modeling and robust variational inference and is equipped with natural uncertainty estimates, which will help derive reproducible and accurate prediction for robust decision making, with the ultimate goal of improving human health outcomes, as showcased in the three experiments
Summary
  • Introduction:

    Modern high-throughput molecular profiling technologies have produced rich high-dimensional data for different bio-molecules at the genome, constituting genome, transcriptome, translatome, proteome, metabolome, epigenome, and interactome scales [Huang et al, 2017, Hajiramezanali et al, 2018b, 2019b, Karimi et al, 2020, Pakbin et al, 2018]
  • Such multi-view data span a diverse range of cellular activities, developing an understanding of how these data types quantitatively relate to each other and to phenotypic characteristics remains elusive.
  • The presented work contains three major contributions: 1) The authors propose a novel Bayesian relation learning framework, BayReL, that can flexibly incorporate the available graph dependency structure of each view. 2) It can exploit non-linear transformations and provide probabilistic interpretation simultaneously. 3) It can infer interactions across different heterogeneous features of input datasets, which is critical to derive meaningful biological knowledge for integrative multi-omics data analysis
  • Methods:

    The authors propose a new graph-structured data integration method, Bayesian Relational Learning (BayReL), for integrative analysis of multi-omics data.
  • The goal of the model is to find inter-relations between nodes of the graphs in different views.
  • The authors model these relations as edges of a multi-partite graph G.
  • The nodes in the multi-partite graph G are the union of the nodes in all views, i.e. VG =
  • Results:

    Considering higher than 97% negative accuracy, the best positive accuracy of BayReL, BCCA, and SRCA are 82.7% ± 4.7, 28.30% ± 3.21, and 26.41%, respectively.
  • BayReL substantially outperforms the baselines with up to 54% margin.
  • Using 50% of all the samples, while the average prediction sensitivity of BayReL reduces less than 2% in the worst case scenario, BCCA’s performance degraded around 6%.
  • Following Lee et al [2018], the authors study 53 out of 160 drugs that have less than 50% cell viability in at least half of the patient samples
  • Conclusion:

    The authors have proposed BayReL, a novel Bayesian relational representation learning method that infers interactions across multi-omics data types.
  • BayReL is unique in its model and potential applications
  • This novel generative model is able to deal with growing complexity and heterogeneity of modern large-scale data with complex dependency structures, which is especially critical when analyzing multi-omics data to derive biological insights, the main focus of the research.
  • BayReL is a generative model with Bayesian modeling and robust variational inference and is equipped with natural uncertainty estimates, which will help derive reproducible and accurate prediction for robust decision making, with the ultimate goal of improving human health outcomes, as showcased in the three experiments
Tables
  • Table1: Comparison of prediction sensitivity (in %) in TCGA for different graph densities
  • Table2: Prediction sensitivity (in %) in TCGA for different percentage of training samples
  • Table3: Comparison of prediction sensitivity (in %) in AML dataset for different graph densities
  • Table4: Enriched GO terms for the top 200 interactions in AML data
Download tables as Excel
Related work
  • Graph-regularized CCA (gCCA). There are several recent CCA extensions that learn shared lowdimensional representations of multiple sources using the graph-induced knowledge of common sources [Chen et al, 2019a, 2018]. They directly impose the dependency graph between samples into a regularizer term, but are not capable of considering the dependency graph between features. These methods are closely related to classic graph-aware regularizers for dimension reduction [Jiang et al, 2013], data reconstruction, clustering [Shang et al, 2012], and classification. Similar to classical CCA methods, they cannot cope with high-dimensional data of small sample sizes while multi-omics data is typically that way when studying complex disease. In addition, these methods focus on latent representation learning but do not explicitly model relational dependency between features across views. Hence, they often require ad-hoc post-processing steps, such as taking correlation and thresholding, to infer inter-relations. Bayesian CCA. Beyond classical linear algebraic solution based CCA methods, there is a rich literature on generative modelling interpretation of CCA [Bach and Jordan, 2005, Virtanen et al, 2011, Klami et al, 2013]. These methods are attractive for their hierarchical construction, improving their interpretability and expressive power, as well as dealing with high dimensional data of small sample size. Some of them, such as [Bach and Jordan, 2005, Klami et al, 2013], are generic factor analysis models that decompose the data into shared and view-specific components and include an additional constraint to extract the statistical dependencies between views. Most of the generative methods retain the linear nature of CCA, but provide inference methods that are more robust than the classical solution. There are also a number of recent variational autoencoder based models that incorporate non-linearity in addition to having the probabilistic interpretability of CCA [Virtanen et al, 2011, Gundersen et al, 2019]. Our BayReL is similar as these methods in allowing non-linear transformations. However, these models attempt to learn low-dimensional latent variables for multiple views while the focus of BayReL is to take advantage of a priori known relationships among features of the same type, modeled as a graph at each corresponding view, to infer a multi-partite graph that encodes the interactions across views. Link prediction. In recent years, several graph neural network architectures have been shown to be effective for link prediction by low-dimensional embedding [Hamilton et al, 2017, Kipf and Welling, 2016, Hasanzadeh et al, 2019]. The majority of these methods do not incorporate heterogeneous graphs, with multiple types of nodes and edges, or graphs with heterogeneous node attributes [Zhang et al, 2019]. In this paper, we have to deal with multiple types of nodes, edges, and attributes in multi-omics data integration. The node embedding of our model is closely related to the Variational Graph AutoEncoder (VGAE) introduced by Kipf and Welling [2016]. However, the original VGAE is designed for node embedding in a single homogeneous graph setting while in our model we learn node embedding for all views. Furthermore, our model can be used for prediction of missing edges in each specific view. BayReL can also be adopted for graph transfer learning between two heterogeneous views to improve the link prediction in each view instead of learning them separately. We leave this for future study. Geometric matrix completion. There have been attempts to incorporate graph structure in matrix completion for recommender systems [Berg et al, 2017, Monti et al, 2017, Kalofolias et al, 2014, Ma et al, 2011, Hasanzadeh et al, 2019]. These methods take advantage of the known item-item and user-user relationships and their attributes to complete the user-item rating matrix. These methods either add a graph-based regularizer [Kalofolias et al, 2014, Ma et al, 2011], or use graph neural networks [Monti et al, 2017] in their analyses. Our method is closely related to the latter one. However, all of these methods assume that the matrix (i.e. inter-relations) is partially observed while we do not require such an assumption in BayReL, which is inherent advantage of formulating the problem as a generative model. In most of existing integrative multi-omics data analyses, there are no a priori known inter-relations.
Funding
  • Acknowledgments and Disclosure of Funding The presented materials are based upon the work supported by the National Science Foundation under Grants IIS-1848596, CCF-1553281, IIS-1812641, ECCS-1839816, and CCF-1934904
Study subjects and analysis
real-world datasets: 3
In most of existing integrative multi-omics data analyses, there are no a priori known inter-relations. We test the performance of BayReL on capturing meaningful inter-relations across views on three real-world datasets. We compare our model with two baselines, Bayesian CCA (BCCA) [Klami et al, 2013] and Spearman’s Rank Correlation Analysis (SRCA) of raw datasets

patients: 172
While anaerobes dominate in low oxygen and low pH environments, pathogens, in particular P. aeruginosa, dominate in the opposite conditions [Morton et al, 2019]. The dataset includes 16S ribosomal RNA (rRNA) sequencing and metabolomics for 172 patients with g_Pseudomonas g_Pseudomonas|s_veronii g_Pseudomonas|s_fragi Anaerobic microbes. 0.4 0.25Posi0ti.v5e0Acc0u.7r5acy 1.00

samples: 10
CF. Following Morton et al [2019], we filter out microbes that appear in less than ten samples, due to the overwhelming sparsity of microbiome data, resulting in 138 unique microbial taxa and 462 metabolite features. We use the reported target molecules of P. aeruginosa in studies Quinn et al [2015] and Morton et al [2019] as a validation set for the microbiome-metabolome interactions

genes: 11872
The TCGA data contains both miRNA and gene expression data for 1156 breast cancer (BRCA) tumor patients. For RNA-Seq data, we filter out the genes with low expression, requiring each gene to have at least 10 count per million in at least 25% of the samples, resulting in 11872 genes for our analysis. We further remove the sequencing depth effect using edgeR [Robinson et al, 2010]

AML patients: 30
Data description. This in vitro drug sensitivity study has both gene expression and drug sensitivity data to a panel of 160 chemotherapy drugs and targeted inhibitors across 30 AML patients [Lee et al, 2018]. While 62 drugs are approved by the U.S Food and Drug Administration (FDA) and encompassed a broad range of drug action mechanisms, the others are investigational drugs for cancer patients

genes: 9073
Similar at the Cancer Cell Line Encyclopedia (CCLE) [Barretina et al, 2012] and MERGE [Lee et al, 2018] studies, we use the area under the curve (AUC) to indicate drug sensitivity across a range of drug concentrations. For gene expression, we pre-processed RNA-Seq data for 9073 genes [Lee et al, 2018]. Experimental details and evaluation metrics

patients: 30
Note that DGIdb contains only the interactions for 43 of the 53 drugs included in our study. 2) Consistency of significant gene-drug interactions in two different AML datasets with 30 patients and 14 cell lines. We compare BayReL with BCCA in consistency of significant gene-drug interactions, where all 30 patient samples are used for discovery and the discovered interactions are validated using 14 cell lines

patient samples: 30
2) Consistency of significant gene-drug interactions in two different AML datasets with 30 patients and 14 cell lines. We compare BayReL with BCCA in consistency of significant gene-drug interactions, where all 30 patient samples are used for discovery and the discovered interactions are validated using 14 cell lines. Numerical results

patient samples: 30
Avg. degree SRCA BCCA BayReL. We also compare the gene-drug interactions when we learn the graph using all 30 patient samples and 14 cell lines. The KL divergence between two inferred bipartite graphs are 0.38 and 0.66 for BayReL and BCCA, respectively

Reference
  • Martín Abadi, Ashish Agarwal, Paul Barham, Eugene Brevdo, Zhifeng Chen, Craig Citro, Greg S. Corrado, Andy Davis, Jeffrey Dean, Matthieu Devin, Sanjay Ghemawat, Ian Goodfellow, Andrew Harp, Geoffrey Irving, Michael Isard, Yangqing Jia, Rafal Jozefowicz, Lukasz Kaiser, Manjunath Kudlur, Josh Levenberg, Dandelion Mané, Rajat Monga, Sherry Moore, Derek Murray, Chris Olah, Mike Schuster, Jonathon Shlens, Benoit Steiner, Ilya Sutskever, Kunal Talwar, Paul Tucker, Vincent Vanhoucke, Vijay Vasudevan, Fernanda Viégas, Oriol Vinyals, Pete Warden, Martin Wattenberg, Martin Wicke, Yuan Yu, and Xiaoqiang Zheng. TensorFlow: Large-scale machine learning on heterogeneous systems, 2015. URL https://www.tensorflow.org/. Software available from tensorflow.org.
    Findings
  • Eduardo Andrés-León, Ildefonso Cases, Sergio Alonso, and Ana M Rojas. Novel mirna-mrna interactions conserved in essential cancer pathways. Scientific reports, 7:46101, 2017.
    Google ScholarLocate open access versionFindings
  • Galen Andrew, Raman Arora, Jeff Bilmes, and Karen Livescu. Deep canonical correlation analysis. In International conference on machine learning, pages 1247–1255, 2013.
    Google ScholarLocate open access versionFindings
  • Francis R Bach and Michael I Jordan. A probabilistic interpretation of canonical correlation analysis. 2005.
    Google ScholarFindings
  • Esther B Bächli, Dominik J Schaer, Roland B Walter, Jörg Fehr, and Gabriele Schoedon. Functional expression of the cd163 scavenger receptor on acute myeloid leukemia cells of monocytic lineage. Journal of leukocyte biology, 79(2):312–318, 2006.
    Google ScholarLocate open access versionFindings
  • Jordi Barretina, Giordano Caponigro, Nicolas Stransky, Kavitha Venkatesan, Adam A Margolin, Sungjoon Kim, Christopher J Wilson, Joseph Lehár, Gregory V Kryukov, Dmitriy Sonkin, et al. The cancer cell line encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature, 483(7391):603–607, 2012.
    Google ScholarLocate open access versionFindings
  • Yoav Benjamini and Yosef Hochberg. Controlling the false discovery rate: a practical and powerful approach to multiple testing. Journal of the Royal statistical society: series B (Methodological), 57(1):289–300, 1995.
    Google ScholarLocate open access versionFindings
  • Rianne van den Berg, Thomas N Kipf, and Max Welling. Graph convolutional matrix completion. arXiv preprint arXiv:1706.02263, 2017.
    Findings
  • Annamaria Bevivino, Giovanni Bacci, Pavel Drevinek, Maria T Nelson, Lucas Hoffman, and Alessio Mengoni. Deciphering the ecology of cystic fibrosis bacterial communities: towards systems-level integration. Trends in molecular medicine, 2019.
    Google ScholarLocate open access versionFindings
  • DG Burke, Fiona Fouhy, MJ Harrison, Mary C Rea, Paul D Cotter, Orla O’Sullivan, Catherine Stanton, C Hill, Fergus Shanahan, Barry J Plant, et al. The altered gut microbiota in adults with cystic fibrosis. BMC microbiology, 17(1):58, 2017.
    Google ScholarLocate open access versionFindings
  • Jia Chen, Gang Wang, Yanning Shen, and Georgios B Giannakis. Canonical correlation analysis of datasets with a common source graph. IEEE Transactions on Signal Processing, 66(16):4398–4408, 2018.
    Google ScholarLocate open access versionFindings
  • Jia Chen, Gang Wang, and Georgios B Giannakis. Multiview canonical correlation analysis over graphs. In ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 2947–2951. IEEE, 2019a.
    Google ScholarLocate open access versionFindings
  • Yang-Yi Chen, Wei-An Chang, En-Shyh Lin, Yi-Jen Chen, and Po-Lin Kuo. Expressions of hla class ii genes in cutaneous melanoma were associated with clinical outcome: Bioinformatics approaches and systematic analysis of public microarray and rna-seq datasets. Diagnostics, 9(2):59, 2019b.
    Google ScholarLocate open access versionFindings
  • Matthew J Christopher, Allegra A Petti, Michael P Rettig, Christopher A Miller, Ezhilarasi Chendamarai, Eric J Duncavage, Jeffery M Klco, Nicole M Helton, Michelle O’Laughlin, Catrina C Fronick, et al. Immune escape of relapsed aml cells after allogeneic transplantation. New England Journal of Medicine, 379(24):2330–2341, 2018.
    Google ScholarLocate open access versionFindings
  • Michael Defferrard, Xavier Bresson, and Pierre Vandergheynst. Convolutional neural networks on graphs with fast localized spectral filtering. In Advances in Neural Information Processing Systems, pages 3844–3852, 2016.
    Google ScholarLocate open access versionFindings
  • Yarin Gal, Jiri Hron, and Alex Kendall. Concrete dropout. In Advances in neural information processing systems, pages 3581–3590, 2017.
    Google ScholarLocate open access versionFindings
  • Zoubin Ghahramani. Probabilistic machine learning and artificial intelligence. Nature, 521(7553): 452–459, 2015.
    Google ScholarLocate open access versionFindings
  • Gregory Gundersen, Bianca Dumitrascu, Jordan T Ash, and Barbara E Engelhardt. End-to-end training of deep probabilistic cca on paired biomedical observations. In 35th Conference on Uncertainty in Artificial Intelligence, UAI 2019, 2019.
    Google ScholarLocate open access versionFindings
  • Ehsan Hajiramezanali, Siamak Zamani Dadaneh, Paul de Figueiredo, Sing-Hoi Sze, Mingyuan Zhou, and Xiaoning Qian. Differential expression analysis of dynamical sequencing count data with a gamma Markov chain. arXiv preprint arXiv:1803.02527, 2018a.
    Findings
  • Ehsan Hajiramezanali, Siamak Zamani Dadaneh, Alireza Karbalayghareh, Mingyuan Zhou, and Xiaoning Qian. Bayesian multi-domain learning for cancer subtype discovery from next-generation sequencing count data. In Advances in Neural Information Processing Systems, pages 9115–9124, 2018b.
    Google ScholarLocate open access versionFindings
  • Ehsan Hajiramezanali, Arman Hasanzadeh, Nick Duffield, Krishna R Narayanan, Mingyuan Zhou, and Xiaoning Qian. Variational graph recurrent neural networks. In Advances in Neural Information Processing Systems, 2019a.
    Google ScholarLocate open access versionFindings
  • Ehsan Hajiramezanali, Mahdi Imani, Ulisses Braga-Neto, Xiaoning Qian, and Edward R Dougherty. Scalable optimal bayesian classification of single-cell trajectories under regulatory model uncertainty. BMC genomics, 20(6):435, 2019b.
    Google ScholarLocate open access versionFindings
  • Ehsan Hajiramezanali, Arman Hasanzadeh, Nick Duffield, Krishna Narayanan, and Xiaoning Qian. Semi-implicit stochastic recurrent neural networks. In ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 3342–3346, 2020.
    Google ScholarLocate open access versionFindings
  • Will Hamilton, Zhitao Ying, and Jure Leskovec. Inductive representation learning on large graphs. In Advances in neural information processing systems, pages 1024–1034, 2017.
    Google ScholarLocate open access versionFindings
  • A. Hasanzadeh, X. Liu, N. Duffield, and K. R. Narayanan. Piecewise stationary modeling of random processes over graphs with an application to traffic prediction. In 2019 IEEE International Conference on Big Data (Big Data), pages 3779–3788, 2019.
    Google ScholarLocate open access versionFindings
  • Arman Hasanzadeh, Ehsan Hajiramezanali, Nick Duffield, Krishna R Narayanan, Mingyuan Zhou, and Xiaoning Qian. Semi-implicit graph variational auto-encoders. In Advances in Neural Information Processing Systems, 2019.
    Google ScholarLocate open access versionFindings
  • Arman Hasanzadeh, Ehsan Hajiramezanali, Shahin Boluki, Mingyuan Zhou, Nick Duffield, Krishna Narayanan, and Xiaoning Qian. Bayesian graph neural networks with adaptive connection sampling. arXiv preprint arXiv:2006.04064, 2020.
    Findings
  • Sijia Huang, Kumardeep Chaudhary, and Lana X Garmire. More is better: recent progress in multi-omics data integration methods. Frontiers in genetics, 8:84, 2017.
    Google ScholarLocate open access versionFindings
  • Zepeng Huo, Arash PakBin, Xiaohan Chen, Nathan Hurley, Ye Yuan, Xiaoning Qian, Zhangyang Wang, Shuai Huang, and Bobak Mortazavi. Uncertainty quantification for deep context-aware mobile activity recognition and unknown context discovery. arXiv preprint arXiv:2003.01753, 2020.
    Findings
  • Bo Jiang, Chris Ding, Bio Luo, and Jin Tang. Graph-laplacian pca: Closed-form solution and robustness. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 3492–3498, 2013.
    Google ScholarLocate open access versionFindings
  • Vassilis Kalofolias, Xavier Bresson, Michael Bronstein, and Pierre Vandergheynst. Matrix completion on graphs. arXiv preprint arXiv:1408.1717, 2014.
    Findings
  • Mostafa Karimi, Arman Hasanzadeh, and Yang Shen. Network-principled deep generative models for designing drug combinations as graph sets. Bioinformatics, 36(Supplement_1):i445–i454, 07 2020. ISSN 1367-4803.
    Google ScholarLocate open access versionFindings
  • Thomas N Kipf and Max Welling. Variational graph auto-encoders. arXiv preprint arXiv:1611.07308, 2016.
    Findings
  • Thomas N Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. In International Conference on Learning Representations, 2017.
    Google ScholarLocate open access versionFindings
  • Arto Klami, Seppo Virtanen, and Samuel Kaski. Bayesian canonical correlation analysis. Journal of Machine Learning Research, 14(Apr):965–1003, 2013.
    Google ScholarLocate open access versionFindings
  • William Kong, Lili He, Marc Coppola, Jianping Guo, Nicole N Esposito, Domenico Coppola, and Jin Q Cheng. Microrna-155 regulates cell survival, growth, and chemosensitivity by targeting foxo3a in breast cancer. Journal of Biological Chemistry, 285(23):17869–17879, 2010.
    Google ScholarLocate open access versionFindings
  • Su-In Lee, Safiye Celik, Benjamin A Logsdon, Scott M Lundberg, Timothy J Martins, Vivian G Oehler, Elihu H Estey, Chris P Miller, Sylvia Chien, Jin Dai, et al. A machine learning approach to integrate big data for precision medicine in acute myeloid leukemia. Nature communications, 9(1): 1–13, 2018.
    Google ScholarLocate open access versionFindings
  • Jianwei Li, Shan Zhang, Yanping Wan, Yingshu Zhao, Jiangcheng Shi, Yuan Zhou, and Qinghua Cui. Misim v2. 0: a web server for inferring microrna functional similarity based on microrna-disease associations. Nucleic acids research, 47(W1):W536–W541, 2019.
    Google ScholarLocate open access versionFindings
  • Ruediger Liersch, Joachim Gerss, Christoph Schliemann, Michael Bayer, Christian Schwöppe, Christoph Biermann, Iris Appelmann, Torsten Kessler, Bob Löwenberg, Thomas Büchner, et al. Osteopontin is a prognostic factor for survival of acute myeloid leukemia patients. Blood, The Journal of the American Society of Hematology, 119(22):5215–5220, 2012.
    Google ScholarLocate open access versionFindings
  • Hao Ma, Dengyong Zhou, Chao Liu, Michael R Lyu, and Irwin King. Recommender systems with social regularization. In Proceedings of the fourth ACM international conference on Web search and data mining, pages 287–296, 2011.
    Google ScholarLocate open access versionFindings
  • X Meng, J Wang, C Yuan, X Li, Y Zhou, Ralf Hofestädt, and Ming Chen. Cancernet: a database for decoding multilevel molecular interactions across diverse cancer types. Oncogenesis, 4(12): e177–e177, 2015.
    Google ScholarLocate open access versionFindings
  • Francesco Miragoli, Sara Federici, Susanna Ferrari, Andrea Minuti, Annalisa Rebecchi, Eugenia Bruzzese, Vittoria Buccigrossi, Alfredo Guarino, and Maria Luisa Callegari. Impact of cystic fibrosis disease on archaea and bacteria composition of gut microbiota. FEMS microbiology ecology, 93(2):fiw230, 2017.
    Google ScholarLocate open access versionFindings
  • Federico Monti, Michael Bronstein, and Xavier Bresson. Geometric matrix completion with recurrent multi-graph neural networks. In Advances in Neural Information Processing Systems, pages 3697–3707, 2017.
    Google ScholarLocate open access versionFindings
  • James T Morton, Alexander A Aksenov, Louis Felix Nothias, James R Foulds, Robert A Quinn, Michelle H Badri, Tami L Swenson, Marc W Van Goethem, Trent R Northen, Yoshiki VazquezBaeza, et al. Learning representations of microbe–metabolite interactions. Nature methods, 16 (12):1306–1314, 2019.
    Google ScholarLocate open access versionFindings
  • Arash Pakbin, Parvez Rafi, Nate Hurley, Wade Schulz, M Harlan Krumholz, and J Bobak Mortazavi. Prediction of icu readmissions using data at patient discharge. In 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pages 4932–4935. IEEE, 2018.
    Google ScholarLocate open access versionFindings
  • Robert A Quinn, Katrine Whiteson, Yan-Wei Lim, Peter Salamon, Barbara Bailey, Simone Mienardi, Savannah E Sanchez, Don Blake, Doug Conrad, and Forest Rohwer. A winogradsky-based culture system shows an association between microbial fermentation and cystic fibrosis exacerbation. The ISME journal, 9(4):1024–1038, 2015.
    Google ScholarLocate open access versionFindings
  • Mark D Robinson, Davis J McCarthy, and Gordon K Smyth. edger: a bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics, 26(1):139–140, 2010.
    Google ScholarLocate open access versionFindings
  • Fanhua Shang, LC Jiao, and Fei Wang. Graph dual regularization non-negative matrix factorization for co-clustering. Pattern Recognition, 45(6):2237–2250, 2012.
    Google ScholarLocate open access versionFindings
  • Jie Shen, Qiang Hu, Michael Schrauder, Li Yan, Dan Wang, Leonardo Medico, Yuqing Guo, Song Yao, Qianqian Zhu, Biao Liu, et al. Circulating mir-148b and mir-133a as biomarkers for breast cancer detection. Oncotarget, 5(14):5284, 2014.
    Google ScholarLocate open access versionFindings
  • Daša Jevšinek Skok, Nina Hauptman, Emanuela Boštjancic, and Nina Zidar. The integrative knowledge base for mirna-mrna expression in colorectal cancer. Scientific Reports, 9(1):1–9, 2019.
    Google ScholarLocate open access versionFindings
  • [1] Esther B Bächli, Dominik J Schaer, Roland B Walter, Jörg Fehr, and Gabriele Schoedon. Functional expression of the cd163 scavenger receptor on acute myeloid leukemia cells of monocytic lineage. Journal of leukocyte biology, 79(2):312–318, 2006.
    Google ScholarLocate open access versionFindings
  • [2] Yang-Yi Chen, Wei-An Chang, En-Shyh Lin, Yi-Jen Chen, and Po-Lin Kuo. Expressions of hla class ii genes in cutaneous melanoma were associated with clinical outcome: Bioinformatics approaches and systematic analysis of public microarray and rna-seq datasets. Diagnostics, 9(2): 59, 2019.
    Google ScholarLocate open access versionFindings
  • [3] Matthew J Christopher, Allegra A Petti, Michael P Rettig, Christopher A Miller, Ezhilarasi Chendamarai, Eric J Duncavage, Jeffery M Klco, Nicole M Helton, Michelle O’Laughlin, Catrina C Fronick, et al. Immune escape of relapsed aml cells after allogeneic transplantation. New England Journal of Medicine, 379(24):2330–2341, 2018.
    Google ScholarLocate open access versionFindings
  • [4] Ruediger Liersch, Joachim Gerss, Christoph Schliemann, Michael Bayer, Christian Schwöppe, Christoph Biermann, Iris Appelmann, Torsten Kessler, Bob Löwenberg, Thomas Büchner, et al. Osteopontin is a prognostic factor for survival of acute myeloid leukemia patients. Blood, The Journal of the American Society of Hematology, 119(22):5215–5220, 2012.
    Google ScholarLocate open access versionFindings
Your rating :
0

 

Tags
Comments