Xiaodong Cai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Xiaodong Cai has a total of 48 co-authors. Here are the top co-authors by number of shared publications:
Xin Zhou
Department of Electrical and Computer Engineering, University of Miami , Coral Gables, FL 33146, USA
5 shared publicationsJi Wen
Department of Electrical and Computer Engineering, University of Miami, Coral Gables, USA
4 shared publicationsG.B. Giannakis
3 shared publications- Georgios B. Giannakis
Department of Electrical and Computer Engineering, University of Minnesota, Minneapolis, MN, USA
3 shared publications - Shizhong Xu
Department of Botany and Plant Sciences, University of California, Riverside, USA
3 shared publications Anhui Huang
Department of Electrical and Computer Engineering, University of Miami, Coral Gables, Florida, United States of America
2 shared publications
Latest Publications
- 29 October 2026
- Published by BMJ in British Journal of Ophthalmology
Deep learning for identifying Alzheimer’s disease and mild cognitive impairment with optical coherence tomography angiography
Background/aims To develop a machine learning (ML) method with deep learning (DL) models for detecting Alzheimer’s disease (AD) and mild cognitive impairment (MCI) using optical coherence tomography angiography (OCTA) images. Methods OCTA images of 173 eyes of 129 subjects with AD (40 eyes, 28 subjects), MCI (62 eyes, 36 subjects) or cognitively normal (CN) (71 eyes, 65 subjects older than 49 years) were acquired. Two pretrained DL models, DINO and ResNet50, were used to extract features from OCTA images of the superficial vascular plexus and the deep vascular plexus. The extracted features were used to train three ML models, support vector machine (SVM), XGBoost and logistic regression with elastic net regularisation (LR-EN), to predict AD or MCI. For performance comparison, convolutional neural networks were also trained to predict AD or MCI. In addition, 14 biomarkers such as vessel density and foveal avascular zone area were extracted from OCTA images and used to train SVM, XGBoost and LR-EN. Results SVM with DINO features achieved the best performance with an area under the receiver operating characteristic curve (AUC) of 0.820, 0.756 and 0.695 in binary classification for discriminating AD versus CN, MCI versus CN and AD versus MCI, respectively. It also achieved the best average AUC of 0.741 in three-class classification for distinguishing AD, MCI and CN. Conclusion Our ML method achieved significantly better prediction accuracy than existing ML methods. Moreover, since our method trains classifiers using features extracted from OCTA images, it enables multiple organisations to collaborate on training an ML/DL model without sharing raw data, which can significantly increase the sample size of the training data, thereby improving the prediction accuracy of the trained model.
- 29 October 2026
- Published by BMJ in British Journal of Ophthalmology
Deep learning for identifying Alzheimer’s disease and mild cognitive impairment with optical coherence tomography angiography
Background/aims To develop a machine learning (ML) method with deep learning (DL) models for detecting Alzheimer’s disease (AD) and mild cognitive impairment (MCI) using optical coherence tomography angiography (OCTA) images. Methods OCTA images of 173 eyes of 129 subjects with AD (40 eyes, 28 subjects), MCI (62 eyes, 36 subjects) or cognitively normal (CN) (71 eyes, 65 subjects older than 49 years) were acquired. Two pretrained DL models, DINO and ResNet50, were used to extract features from OCTA images of the superficial vascular plexus and the deep vascular plexus. The extracted features were used to train three ML models, support vector machine (SVM), XGBoost and logistic regression with elastic net regularisation (LR-EN), to predict AD or MCI. For performance comparison, convolutional neural networks were also trained to predict AD or MCI. In addition, 14 biomarkers such as vessel density and foveal avascular zone area were extracted from OCTA images and used to train SVM, XGBoost and LR-EN. Results SVM with DINO features achieved the best performance with an area under the receiver operating characteristic curve (AUC) of 0.820, 0.756 and 0.695 in binary classification for discriminating AD versus CN, MCI versus CN and AD versus MCI, respectively. It also achieved the best average AUC of 0.741 in three-class classification for distinguishing AD, MCI and CN. Conclusion Our ML method achieved significantly better prediction accuracy than existing ML methods. Moreover, since our method trains classifiers using features extracted from OCTA images, it enables multiple organisations to collaborate on training an ML/DL model without sharing raw data, which can significantly increase the sample size of the training data, thereby improving the prediction accuracy of the trained model.
- Open Access
Deep Semi-Supervised Survival Analysis for Predicting Cancer Prognosis
The Cox Proportional Hazards (PH) model is widely used in survival analysis. Recently, artificial neural network (ANN)-based Cox-PH models have been developed. However, training these Cox models with high-dimensional features typically requires a substantial number of labeled samples containing information about time-to-event. The limited availability of labeled data for training often constrains the performance of ANN-based Cox models. To address this issue, we employed a deep semi-supervised learning (DSSL) approach to develop single- and multi-modal ANN-based Cox models based on the Mean Teacher (MT) framework, which utilizes both labeled and unlabeled data for training. We applied our model, named Cox-MT, to predict the prognosis of several types of cancer using data from The Cancer Genome Atlas (TCGA). Our single-modal Cox-MT models, utilizing TCGA RNA-seq data or whole slide images, significantly outperformed the existing ANN-based Cox model, Cox-nnet, using the same data set across four types of cancer considered. As the number of unlabeled samples increased, the performance of Cox-MT significantly improved with a given set of labeled data. Furthermore, our multi-modal Cox-MT model demonstrated considerably better performance than the single-modal model. In summary, the Cox-MT model effectively leverages both labeled and unlabeled data to significantly enhance prediction accuracy compared to existing ANN-based Cox models trained solely on labeled data.
Open AccessDeep Semi-Supervised Survival Analysis for Predicting Cancer Prognosis
The Cox Proportional Hazards (PH) model is widely used in survival analysis. Recently, artificial neural network (ANN)-based Cox-PH models have been developed. However, training these Cox models with high-dimensional features typically requires a substantial number of labeled samples containing information about time-to-event. The limited availability of labeled data for training often constrains the performance of ANN-based Cox models. To address this issue, we employed a deep semi-supervised learning (DSSL) approach to develop single- and multi-modal ANN-based Cox models based on the Mean Teacher (MT) framework, which utilizes both labeled and unlabeled data for training. We applied our model, named Cox-MT, to predict the prognosis of several types of cancer using data from The Cancer Genome Atlas (TCGA). Our single-modal Cox-MT models, utilizing TCGA RNA-seq data or whole slide images, significantly outperformed the existing ANN-based Cox model, Cox-nnet, using the same data set across four types of cancer considered. As the number of unlabeled samples increased, the performance of Cox-MT significantly improved with a given set of labeled data. Furthermore, our multi-modal Cox-MT model demonstrated considerably better performance than the single-modal model. In summary, the Cox-MT model effectively leverages both labeled and unlabeled data to significantly enhance prediction accuracy compared to existing ANN-based Cox models trained solely on labeled data.
- Open Access
- 23 September 2024
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 25 (6)
Deep contrastive learning for predicting cancer prognosis using gene expression values
Recent advancements in image classification have demonstrated that contrastive learning (CL) can aid in further learning tasks by acquiring good feature representation from a limited number of data samples. In this paper, we applied CL to tumor transcriptomes and clinical data to learn feature representations in a low-dimensional space. We then utilized these learned features to train a classifier to categorize tumors into a high- or low-risk group of recurrence. Using data from The Cancer Genome Atlas (TCGA), we demonstrated that CL can significantly improve classification accuracy. Specifically, our CL-based classifiers achieved an area under the receiver operating characteristic curve (AUC) greater than 0.8 for 14 types of cancer, and an AUC greater than 0.9 for 3 types of cancer. We also developed CL-based Cox (CLCox) models for predicting cancer prognosis. Our CLCox models trained with the TCGA data outperformed existing methods significantly in predicting the prognosis of 19 types of cancer under consideration. The performance of CLCox models and CL-based classifiers trained with TCGA lung and prostate cancer data were validated using the data from two independent cohorts. We also show that the CLCox model trained with the whole transcriptome significantly outperforms the Cox model trained with the 16 genes of Oncotype DX that is in clinical use for breast cancer patients. The trained models and the Python codes are publicly accessible and provide a valuable resource that will potentially find clinical applications for many types of cancer.
- 23 September 2024
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 25 (6)
Open AccessDeep contrastive learning for predicting cancer prognosis using gene expression values
Recent advancements in image classification have demonstrated that contrastive learning (CL) can aid in further learning tasks by acquiring good feature representation from a limited number of data samples. In this paper, we applied CL to tumor transcriptomes and clinical data to learn feature representations in a low-dimensional space. We then utilized these learned features to train a classifier to categorize tumors into a high- or low-risk group of recurrence. Using data from The Cancer Genome Atlas (TCGA), we demonstrated that CL can significantly improve classification accuracy. Specifically, our CL-based classifiers achieved an area under the receiver operating characteristic curve (AUC) greater than 0.8 for 14 types of cancer, and an AUC greater than 0.9 for 3 types of cancer. We also developed CL-based Cox (CLCox) models for predicting cancer prognosis. Our CLCox models trained with the TCGA data outperformed existing methods significantly in predicting the prognosis of 19 types of cancer under consideration. The performance of CLCox models and CL-based classifiers trained with TCGA lung and prostate cancer data were validated using the data from two independent cohorts. We also show that the CLCox model trained with the whole transcriptome significantly outperforms the Cox model trained with the 16 genes of Oncotype DX that is in clinical use for breast cancer patients. The trained models and the Python codes are publicly accessible and provide a valuable resource that will potentially find clinical applications for many types of cancer.
- Open Access
- 6 September 2021
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 38 (1) , 149-156
Joint eQTL mapping and inference of gene regulatory network improves power of detecting bothcis- andtrans-eQTLs
Genetic variations of expression quantitative trait loci (eQTLs) play a critical role in influencing complex traits and diseases development. Two main factors that affect the statistical power of detecting eQTLs are: (i) relatively small size of samples available, and (ii) heavy burden of multiple testing due to a very large number of variants to be tested. The later issue is particularly severe when one tries to identify trans-eQTLs that are far away from the genes they influence. If one can exploit co-expressed genes jointly in eQTL-mapping, effective sample size can be increased. Furthermore, using the structure of the gene regulatory network (GRN) may help to identify trans-eQTLs without increasing multiple testing burden. In this article, we use the structure equation model (SEM) to model both GRN and effect of eQTLs on gene expression, and then develop a novel algorithm, named sparse SEM for eQTL mapping (SSEMQ), to conduct joint eQTL mapping and GRN inference. The SEM can exploit co-expressed genes jointly in eQTL mapping and also use GRN to determine trans-eQTLs. Computer simulations demonstrate that our SSEMQ significantly outperforms nine existing eQTL mapping methods. SSEMQ is further used to analyze two real datasets of human breast and whole blood tissues, yielding a number of cis- and trans-eQTLs. R package ssemQr is available at https://github.com/Ivis4ml/ssemQr.git. Supplementary data are available at Bioinformatics online.
- 6 September 2021
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 38 (1) , 149-156
Open AccessJoint eQTL mapping and inference of gene regulatory network improves power of detecting bothcis- andtrans-eQTLs
Genetic variations of expression quantitative trait loci (eQTLs) play a critical role in influencing complex traits and diseases development. Two main factors that affect the statistical power of detecting eQTLs are: (i) relatively small size of samples available, and (ii) heavy burden of multiple testing due to a very large number of variants to be tested. The later issue is particularly severe when one tries to identify trans-eQTLs that are far away from the genes they influence. If one can exploit co-expressed genes jointly in eQTL-mapping, effective sample size can be increased. Furthermore, using the structure of the gene regulatory network (GRN) may help to identify trans-eQTLs without increasing multiple testing burden. In this article, we use the structure equation model (SEM) to model both GRN and effect of eQTLs on gene expression, and then develop a novel algorithm, named sparse SEM for eQTL mapping (SSEMQ), to conduct joint eQTL mapping and GRN inference. The SEM can exploit co-expressed genes jointly in eQTL mapping and also use GRN to determine trans-eQTLs. Computer simulations demonstrate that our SSEMQ significantly outperforms nine existing eQTL mapping methods. SSEMQ is further used to analyze two real datasets of human breast and whole blood tissues, yielding a number of cis- and trans-eQTLs. R package ssemQr is available at https://github.com/Ivis4ml/ssemQr.git. Supplementary data are available at Bioinformatics online.
- Open Access
Predicting Gene Expression from DNA Sequence using Residual Neural Network
It is known that cis -acting DNA motifs play an important role in regulating gene expression. The genome in a cell thus contains the information that not only encodes for the synthesis of proteins but also is necessary for regulating expression of genes. Therefore, the mRNA level of a gene may be predictable from the DNA sequence. Indeed, three deep neural network models were developed recently to predict the mRNA level of a gene directly or indirectly from the DNA sequence around the transcription start side of the gene. In this work, we develop a deep residual network model, named ExpResNet, to predict gene expression directly from DNA sequence. Applying ExpResNet to the GTEx data, we demonstrate that ExpResNet outperforms the three existing models across four tissues tested. Our model may be useful in the investigation of gene regulation.
Open AccessPredicting Gene Expression from DNA Sequence using Residual Neural Network
It is known that cis -acting DNA motifs play an important role in regulating gene expression. The genome in a cell thus contains the information that not only encodes for the synthesis of proteins but also is necessary for regulating expression of genes. Therefore, the mRNA level of a gene may be predictable from the DNA sequence. Indeed, three deep neural network models were developed recently to predict the mRNA level of a gene directly or indirectly from the DNA sequence around the transcription start side of the gene. In this work, we develop a deep residual network model, named ExpResNet, to predict gene expression directly from DNA sequence. Applying ExpResNet to the GTEx data, we demonstrate that ExpResNet outperforms the three existing models across four tissues tested. Our model may be useful in the investigation of gene regulation.
- Open Access
Joint eQTL mapping and Inference of Gene Regulatory Network Improves Power of Detecting both cis - and trans -eQTLs
Motivation: Genetic variations of expression quantitative trait loci (eQTLs) play a critical role in influencing complex traits and diseases development. Two main factors that affect the statistical power of detecting eQTLs are: 1) relatively small size of samples available, and 2) heavy burden of multiple testing due to a very large number of variants to be tested. The later issue is particularly severe when one tries to identify trans -eQTLs that are far away from the genes they influence. If one can exploit co-expressed genes jointly in eQTL-mapping, effective sample size can be increased. Furthermore, using the structure of the gene regulatory network (GRN) may help to identify trans -eQTLs without increasing multiple testing burden. Results: In this paper, we employ the structure equation model (SEM) to model both GRN and effect of eQTLs on gene expression, and then develop a novel algorithm, named sparse SEM, for eQTL mapping (SSEMQ) to conduct joint eQTL mapping and GRN inference. The SEM can exploit co-expressed genes jointly in eQTL mapping and also use GRN to determine trans -eQTLs. Computer simulations demonstrate that our SSEMQ significantly outperforms eight existing eQTL mapping methods. SSEMQ is further employed to analyze a real dataset of human breast tissues, yielding a number of cis - and trans -eQTLs. Availability: R package ssemQr is available on https://github.com/Ivis4ml/ssemQr.git .
Open AccessJoint eQTL mapping and Inference of Gene Regulatory Network Improves Power of Detecting both cis - and trans -eQTLs
Motivation: Genetic variations of expression quantitative trait loci (eQTLs) play a critical role in influencing complex traits and diseases development. Two main factors that affect the statistical power of detecting eQTLs are: 1) relatively small size of samples available, and 2) heavy burden of multiple testing due to a very large number of variants to be tested. The later issue is particularly severe when one tries to identify trans -eQTLs that are far away from the genes they influence. If one can exploit co-expressed genes jointly in eQTL-mapping, effective sample size can be increased. Furthermore, using the structure of the gene regulatory network (GRN) may help to identify trans -eQTLs without increasing multiple testing burden. Results: In this paper, we employ the structure equation model (SEM) to model both GRN and effect of eQTLs on gene expression, and then develop a novel algorithm, named sparse SEM, for eQTL mapping (SSEMQ) to conduct joint eQTL mapping and GRN inference. The SEM can exploit co-expressed genes jointly in eQTL mapping and also use GRN to determine trans -eQTLs. Computer simulations demonstrate that our SSEMQ significantly outperforms eight existing eQTL mapping methods. SSEMQ is further employed to analyze a real dataset of human breast tissues, yielding a number of cis - and trans -eQTLs. Availability: R package ssemQr is available on https://github.com/Ivis4ml/ssemQr.git .
- Open Access
- 2 July 2019
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 36 (1) , 197-204
Inference of differential gene regulatory networks based on gene expression and genetic perturbation data
Gene regulatory networks (GRNs) of the same organism can be different under different conditions, although the overall network structure may be similar. Understanding the difference in GRNs under different conditions is important to understand condition-specific gene regulation. When gene expression and other relevant data under two different conditions are available, they can be used by an existing network inference algorithm to estimate two GRNs separately, and then to identify the difference between the two GRNs. However, such an approach does not exploit the similarity in two GRNs, and may sacrifice inference accuracy. In this paper, we model GRNs with the structural equation model (SEM) that can integrate gene expression and genetic perturbation data, and develop an algorithm named fused sparse SEM (FSSEM), to jointly infer GRNs under two conditions, and then to identify difference of the two GRNs. Computer simulations demonstrate that the FSSEM algorithm outperforms the approaches that estimate two GRNs separately. Analysis of a dataset of lung cancer and another dataset of gastric cancer with FSSEM inferred differential GRNs in cancer versus normal tissues, whose genes with largest network degrees have been reported to be implicated in tumorigenesis. The FSSEM algorithm provides a valuable tool for joint inference of two GRNs and identification of the differential GRN under two conditions. The R package fssemR implementing the FSSEM algorithm is available at https://github.com/Ivis4ml/fssemR.git. It is also available on CRAN. Supplementary data are available at Bioinformatics online.
- 2 July 2019
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 36 (1) , 197-204
Open AccessInference of differential gene regulatory networks based on gene expression and genetic perturbation data
Gene regulatory networks (GRNs) of the same organism can be different under different conditions, although the overall network structure may be similar. Understanding the difference in GRNs under different conditions is important to understand condition-specific gene regulation. When gene expression and other relevant data under two different conditions are available, they can be used by an existing network inference algorithm to estimate two GRNs separately, and then to identify the difference between the two GRNs. However, such an approach does not exploit the similarity in two GRNs, and may sacrifice inference accuracy. In this paper, we model GRNs with the structural equation model (SEM) that can integrate gene expression and genetic perturbation data, and develop an algorithm named fused sparse SEM (FSSEM), to jointly infer GRNs under two conditions, and then to identify difference of the two GRNs. Computer simulations demonstrate that the FSSEM algorithm outperforms the approaches that estimate two GRNs separately. Analysis of a dataset of lung cancer and another dataset of gastric cancer with FSSEM inferred differential GRNs in cancer versus normal tissues, whose genes with largest network degrees have been reported to be implicated in tumorigenesis. The FSSEM algorithm provides a valuable tool for joint inference of two GRNs and identification of the differential GRN under two conditions. The R package fssemR implementing the FSSEM algorithm is available at https://github.com/Ivis4ml/fssemR.git. It is also available on CRAN. Supplementary data are available at Bioinformatics online.
- Open Access
- 2 May 2019
- Published by Springer Nature in BMC Bioinformatics
- Vol. 20 (1) , 224
Efficient proximal gradient algorithm for inference of differential gene networks
Gene networks in living cells can change depending on various conditions such as caused by different environments, tissue types, disease states, and development stages. Identifying the differential changes in gene networks is very important to understand molecular basis of various biological process. While existing algorithms can be used to infer two gene networks separately from gene expression data under two different conditions, and then to identify network changes, such an approach does not exploit the similarity between two gene networks, and it is thus suboptimal. A desirable approach would be clearly to infer two gene networks jointly, which can yield improved estimates of network changes. In this paper, we developed a proximal gradient algorithm for differential network (ProGAdNet) inference, that jointly infers two gene networks under different conditions and then identifies changes in the network structure. Computer simulations demonstrated that our ProGAdNet outperformed existing algorithms in terms of inference accuracy, and was much faster than a similar approach for joint inference of gene networks. Gene expression data of breast tumors and normal tissues in the TCGA database were analyzed with our ProGAdNet, and revealed that 268 genes were involved in the changed network edges. Gene set enrichment analysis identified a significant number of gene sets related to breast cancer or other types of cancer that are enriched in this set of 268 genes. Network analysis of the kidney cancer data in the TCGA database with ProGAdNet also identified a set of genes involved in network changes, and the majority of the top genes identified have been reported in the literature to be implicated in kidney cancer. These results corroborated that the gene sets identified by ProGAdNet were very informative about the cancer disease status. A software package implementing the ProGAdNet, computer simulations, and real data analysis is available as Additional file 1. With its superior performance over existing algorithms, ProGAdNet provides a valuable tool for finding changes in gene networks, which may aid the discovery of gene-gene interactions changed under different conditions.
- 2 May 2019
- Published by Springer Nature in BMC Bioinformatics
- Vol. 20 (1) , 224
Open AccessEfficient proximal gradient algorithm for inference of differential gene networks
Gene networks in living cells can change depending on various conditions such as caused by different environments, tissue types, disease states, and development stages. Identifying the differential changes in gene networks is very important to understand molecular basis of various biological process. While existing algorithms can be used to infer two gene networks separately from gene expression data under two different conditions, and then to identify network changes, such an approach does not exploit the similarity between two gene networks, and it is thus suboptimal. A desirable approach would be clearly to infer two gene networks jointly, which can yield improved estimates of network changes. In this paper, we developed a proximal gradient algorithm for differential network (ProGAdNet) inference, that jointly infers two gene networks under different conditions and then identifies changes in the network structure. Computer simulations demonstrated that our ProGAdNet outperformed existing algorithms in terms of inference accuracy, and was much faster than a similar approach for joint inference of gene networks. Gene expression data of breast tumors and normal tissues in the TCGA database were analyzed with our ProGAdNet, and revealed that 268 genes were involved in the changed network edges. Gene set enrichment analysis identified a significant number of gene sets related to breast cancer or other types of cancer that are enriched in this set of 268 genes. Network analysis of the kidney cancer data in the TCGA database with ProGAdNet also identified a set of genes involved in network changes, and the majority of the top genes identified have been reported in the literature to be implicated in kidney cancer. These results corroborated that the gene sets identified by ProGAdNet were very informative about the cancer disease status. A software package implementing the ProGAdNet, computer simulations, and real data analysis is available as Additional file 1. With its superior performance over existing algorithms, ProGAdNet provides a valuable tool for finding changes in gene networks, which may aid the discovery of gene-gene interactions changed under different conditions.
- Open Access
Inference of Differential Gene Regulatory Networks Based on Gene Expression and Genetic Perturbation Data
Motivation: Gene regulatory networks (GRNs) of the same organism can be different under different conditions, although the overall network structure may be similar. Understanding the difference in GRNs under different conditions is important to understand condition-specific gene regulation. When gene expression and other relevant data under two different conditions are available, they can be used by an existing network inference algorithm to estimate two GRNs separately, and then to identify the difference between the two GRNs. However, such an approach does not exploit the similarity in two GRNs, and may sacrifice inference accuracy. Results: In this paper, we model GRNs with the structural equation model (SEM) that can integrate gene expression and genetic perturbation data, and develop an algorithm named fused sparse SEM (FSSEM), to jointly infer GRNs under two conditions, and then to identify difference of the two GRNs. Computer simulations demonstrate that the FSSEM algorithm outperforms the approach that estimates two GRNs separately. Analysis of a gene expression and SNP dataset of lung cancer and normal lung tissues with FSSEM inferred a GRN largely agree with the known lung GRN reported in the literature, and it identified a differential GRN, whose genes with largest degrees were reported to be implicated in lung cancer. The FSSEM algorithm provides a valuable tool for joint inference of two GRNs and identification of the differential GRN under two conditions. Availability: The software package for the FSSEM algorithm is available at https://github.com/Ivis4ml/FSSEM.git Contact: x.cai@miami.edu
Open AccessInference of Differential Gene Regulatory Networks Based on Gene Expression and Genetic Perturbation Data
Motivation: Gene regulatory networks (GRNs) of the same organism can be different under different conditions, although the overall network structure may be similar. Understanding the difference in GRNs under different conditions is important to understand condition-specific gene regulation. When gene expression and other relevant data under two different conditions are available, they can be used by an existing network inference algorithm to estimate two GRNs separately, and then to identify the difference between the two GRNs. However, such an approach does not exploit the similarity in two GRNs, and may sacrifice inference accuracy. Results: In this paper, we model GRNs with the structural equation model (SEM) that can integrate gene expression and genetic perturbation data, and develop an algorithm named fused sparse SEM (FSSEM), to jointly infer GRNs under two conditions, and then to identify difference of the two GRNs. Computer simulations demonstrate that the FSSEM algorithm outperforms the approach that estimates two GRNs separately. Analysis of a gene expression and SNP dataset of lung cancer and normal lung tissues with FSSEM inferred a GRN largely agree with the known lung GRN reported in the literature, and it identified a differential GRN, whose genes with largest degrees were reported to be implicated in lung cancer. The FSSEM algorithm provides a valuable tool for joint inference of two GRNs and identification of the differential GRN under two conditions. Availability: The software package for the FSSEM algorithm is available at https://github.com/Ivis4ml/FSSEM.git Contact: x.cai@miami.edu
- Open Access
Efficient Proximal Gradient Algorithm for Inference of Differential Gene Networks
Background: Gene networks in living cells can change depending on various conditions such as caused by different environments, tissue types, disease states, and development stages. Identifying the differential changes in gene networks is very important to understand molecular basis of various biological process. While existing algorithms can be used to infer two gene networks separately from gene expression data under two different conditions, and then to identify network changes, such an approach does not exploit the data jointly, and it is thus suboptimal. A desirable approach would be clearly to infer two gene networks jointly, which can yield improved estimates of network changes. Results: In this paper, we developed a proximal gradient algorithm for differential network (ProGAdNet) inference, that jointly infers two gene networks under different conditions and then identifies changes in the network structure. Computer simulations demonstrated that our ProGAdNet outperformed existing algorithms in terms of inference accuracy, and was much faster than a similar approach for joint inference of gene networks. Gene expression data of breast tumors and normal tissues in the TCGA database were analyzed with our ProGAdNet, and revealed that 268 genes were involved in the changed network edges. Gene set enrichment analysis of this set of 268 genes identified a number of gene sets related to breast cancer or other types of cancer, which corroborated the gene set identified by ProGAdNet was very informative about the cancer disease status. A software package implementing the ProGAdNet and computer simulations is available upon request. Conclusion: With its superior performance over existing algorithms, ProGAdNet provides a valuable tool for finding changes in gene networks, which may aid the discovery of gene-gene interactions changed under different conditions.
Open AccessEfficient Proximal Gradient Algorithm for Inference of Differential Gene Networks
Background: Gene networks in living cells can change depending on various conditions such as caused by different environments, tissue types, disease states, and development stages. Identifying the differential changes in gene networks is very important to understand molecular basis of various biological process. While existing algorithms can be used to infer two gene networks separately from gene expression data under two different conditions, and then to identify network changes, such an approach does not exploit the data jointly, and it is thus suboptimal. A desirable approach would be clearly to infer two gene networks jointly, which can yield improved estimates of network changes. Results: In this paper, we developed a proximal gradient algorithm for differential network (ProGAdNet) inference, that jointly infers two gene networks under different conditions and then identifies changes in the network structure. Computer simulations demonstrated that our ProGAdNet outperformed existing algorithms in terms of inference accuracy, and was much faster than a similar approach for joint inference of gene networks. Gene expression data of breast tumors and normal tissues in the TCGA database were analyzed with our ProGAdNet, and revealed that 268 genes were involved in the changed network edges. Gene set enrichment analysis of this set of 268 genes identified a number of gene sets related to breast cancer or other types of cancer, which corroborated the gene set identified by ProGAdNet was very informative about the cancer disease status. A software package implementing the ProGAdNet and computer simulations is available upon request. Conclusion: With its superior performance over existing algorithms, ProGAdNet provides a valuable tool for finding changes in gene networks, which may aid the discovery of gene-gene interactions changed under different conditions.