Xiaodi Yang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Xiaodi Yang has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Ziding Zhang
State Key Laboratory of Animal Biotech Breeding, College of Biological Sciences, China Agricultural University , No. 2 Yuanmingyuan West Road, Haidian District, Beijing 100193
9 shared publications - Stefan Wuchty
Department of Computer Science, University of Miami , Coral Gables, FL 33146
6 shared publications - Shiping Yang
State Key Laboratory of Plant Physiology and Biochemistry, College of Biological Sciences, China Agricultural University, Beijing 100193, China
5 shared publications - Yujun Dong
Department of Hematology, Peking University First Hospital
5 shared publications Jialin Zhu
Department of Hematology, Peking University First Hospital
4 shared publications- Xianyi Lian
State Key Laboratory of Agrobiotechnology, College of Biological Sciences, China Agricultural University, Beijing 100193, China
4 shared publications
Latest Publications
- Open Access
- 4 August 2026
- Published by American Society for Microbiology in Microbiology Spectrum
- Vol. 14 (8) , e0399625
Proteomic analysis identifies pathways related to immune dysregulation in patients with hematologic malignancies after COVID-19 infection
Patients with hematologic malignancies (HMs) are particularly vulnerable to coronavirus disease 2019 (COVID-19) because of underlying immune dysfunction and treatment-related immunosuppression. However, proteomic features associated with different clinical trajectories in this population remain insufficiently characterized. We performed serum proteomic analysis in 40 HM patients with COVID-19 and 15 healthy controls. Compared with controls, HM patients showed impaired immune-related responses during the acute phase of COVID-19. Acute-phase proteomic patterns differed across outcome groups; however, because outcome groups were closely intertwined with initial COVID-19 severity, ICU admission, and systemic illness, and because multivariable adjustment was not performed due to the limited sample size, these patterns should be interpreted as severity- and outcome-associated profiles rather than independent trajectory-specific markers. Fatal cases showed evidence of dysregulated immune activation, whereas patients later classified as having long COVID exhibited broader suppression of immune-related pathways. In addition to immune alterations, pathways related to platelet activation and cardiac-related dysfunction were associated with adverse clinical trajectories. Enzyme-linked immunosorbent assay validation supported the association of selected proteins with outcome groups during acute infection. These findings provide a proteomic overview of COVID-19 in HM patients and offer a basis for future mechanistic studies and larger external validation cohorts. IMPORTANCE: Patients with hematologic malignancies are highly vulnerable to severe coronavirus disease 2019 (COVID-19), acute death, and long COVID due to preexisting immune dysfunction. However, the proteomic signatures linked to adverse clinical trajectories remain poorly understood. Our serum proteomic study identifies distinct acute-phase immune profiles associated with different outcomes: broad immune suppression characterizes long COVID, while dysregulated immune activation is associated with fatal cases. Platelet activation and cardiac-related pathways are also linked to poor outcomes. These findings provide key molecular insights for this high-risk population, supporting future biomarker development, risk stratification, and targeted clinical management. CLINICAL TRIALS: This study is registered with ClinicalTrials.gov as NCT05683353 .
- 4 August 2026
- Published by American Society for Microbiology in Microbiology Spectrum
- Vol. 14 (8) , e0399625
Open AccessProteomic analysis identifies pathways related to immune dysregulation in patients with hematologic malignancies after COVID-19 infection
Patients with hematologic malignancies (HMs) are particularly vulnerable to coronavirus disease 2019 (COVID-19) because of underlying immune dysfunction and treatment-related immunosuppression. However, proteomic features associated with different clinical trajectories in this population remain insufficiently characterized. We performed serum proteomic analysis in 40 HM patients with COVID-19 and 15 healthy controls. Compared with controls, HM patients showed impaired immune-related responses during the acute phase of COVID-19. Acute-phase proteomic patterns differed across outcome groups; however, because outcome groups were closely intertwined with initial COVID-19 severity, ICU admission, and systemic illness, and because multivariable adjustment was not performed due to the limited sample size, these patterns should be interpreted as severity- and outcome-associated profiles rather than independent trajectory-specific markers. Fatal cases showed evidence of dysregulated immune activation, whereas patients later classified as having long COVID exhibited broader suppression of immune-related pathways. In addition to immune alterations, pathways related to platelet activation and cardiac-related dysfunction were associated with adverse clinical trajectories. Enzyme-linked immunosorbent assay validation supported the association of selected proteins with outcome groups during acute infection. These findings provide a proteomic overview of COVID-19 in HM patients and offer a basis for future mechanistic studies and larger external validation cohorts. IMPORTANCE: Patients with hematologic malignancies are highly vulnerable to severe coronavirus disease 2019 (COVID-19), acute death, and long COVID due to preexisting immune dysfunction. However, the proteomic signatures linked to adverse clinical trajectories remain poorly understood. Our serum proteomic study identifies distinct acute-phase immune profiles associated with different outcomes: broad immune suppression characterizes long COVID, while dysregulated immune activation is associated with fatal cases. Platelet activation and cardiac-related pathways are also linked to poor outcomes. These findings provide key molecular insights for this high-risk population, supporting future biomarker development, risk stratification, and targeted clinical management. CLINICAL TRIALS: This study is registered with ClinicalTrials.gov as NCT05683353 .
- Open Access
- 11 September 2025
- Published by Elsevier in Molecular & Cellular Proteomics
- Vol. 24 (10) , 101068
Multi-Omics Blood Atlas Reveals Host Immune Response Features of Immunocompromised Populations Following SARS-CoV-2 Infection
- 11 September 2025
- Published by Elsevier in Molecular & Cellular Proteomics
- Vol. 24 (10) , 101068
Open AccessMulti-Omics Blood Atlas Reveals Host Immune Response Features of Immunocompromised Populations Following SARS-CoV-2 Infection
- Open Access
- 31 August 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (5)
Graph neural network integrated with pretrained protein language model for predicting human–virus protein–protein interactions
The systematic identification of human-virus protein–protein interactions (PPIs) is a critical step toward elucidating the underlying mechanisms of viral infection, directly informing the development of targeted interventions against existing and emerging viral threats. In this work, we presented DeepGNHV, an end-to-end framework that integrated a pretrained protein language model with structural features derived from AlphaFold2 and leveraged graph attention networks to predict human-virus PPIs. In comparison to other state-of-the-art approaches, DeepGNHV exhibited superior predictive performance, especially when applied to viral proteins absent from the training process, indicating its strong generalization capability for detecting newly emerging virus-related PPIs. We further demonstrated DeepGNHV’s robustness across diverse perturbations and its practical application under high-confidence thresholds. Additionally, we conducted extensive predictions of human-HPV PPIs, which were supported by multiple lines of evidence and identified several host factors that specifically interact with high-risk HPV. To further explore the biological significance of DeepGNHV, we provided a case study to pinpoint specific residues that play critical roles in facilitating the corresponding PPIs. The source code of DeepGNHV and related data is publicly available on GitHub (https://github.com/bioboy0415/DeepGNHV).
- 31 August 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (5)
Open AccessGraph neural network integrated with pretrained protein language model for predicting human–virus protein–protein interactions
The systematic identification of human-virus protein–protein interactions (PPIs) is a critical step toward elucidating the underlying mechanisms of viral infection, directly informing the development of targeted interventions against existing and emerging viral threats. In this work, we presented DeepGNHV, an end-to-end framework that integrated a pretrained protein language model with structural features derived from AlphaFold2 and leveraged graph attention networks to predict human-virus PPIs. In comparison to other state-of-the-art approaches, DeepGNHV exhibited superior predictive performance, especially when applied to viral proteins absent from the training process, indicating its strong generalization capability for detecting newly emerging virus-related PPIs. We further demonstrated DeepGNHV’s robustness across diverse perturbations and its practical application under high-confidence thresholds. Additionally, we conducted extensive predictions of human-HPV PPIs, which were supported by multiple lines of evidence and identified several host factors that specifically interact with high-risk HPV. To further explore the biological significance of DeepGNHV, we provided a case study to pinpoint specific residues that play critical roles in facilitating the corresponding PPIs. The source code of DeepGNHV and related data is publicly available on GitHub (https://github.com/bioboy0415/DeepGNHV).
- Open Access
- 6 December 2024
- Published by Springer Nature in Communications Biology
- Vol. 7 (1) , 1-12
Genomic differences of patients with hematologic malignancies in different age groups
Hematologic malignancies cause significant morbidity/mortality in both children and young adults (CYAs) as well as older adults (OAs). Yet their biological underpinnings remain inadequately understood. Here, we analyzed clinical and genomic disparities between CYAs and OAs in various hematologic malignancies. We found substantial differences in clinical features such as patient sex, ethnicity, metastasis rates, and tumor subtypes. Genomically, most CYA hematologic malignancies indicated lower mutational burden. Subsequently, we identified differentially mutated genes (DMGs) with varying mutation rates between CYAs and OAs, noting fewer mutations in CYAs for most genes such as TP53, TET2, and DNMT3A. In contrast, several DMGs (i.e., NRAS, KRAS, SMARCA4, ID3, PTPN11, WT1, and KIT) were overrepresented in CYAs. We further investigated human protein interacting partners of these identified DMGs that were highly mutated in CYAs/OAs, respectively, and found significant differences in network topological and functional roles. Notably, CYA malignancies demonstrated extensive copy number alterations (CNAs) and more driver gene fusions. In particular, four CNA differential genes (i.e., ARID1B, MYB, TP53, and ESR1) were overrepresented as amplifications and deletions in CYAs and OAs, respectively. Ultimately, we demonstrated a landscape comparative view of clinically actionable genetic events in CYAs and OAs, providing clues for age-related personalized treatment.
- 6 December 2024
- Published by Springer Nature in Communications Biology
- Vol. 7 (1) , 1-12
Open AccessGenomic differences of patients with hematologic malignancies in different age groups
Hematologic malignancies cause significant morbidity/mortality in both children and young adults (CYAs) as well as older adults (OAs). Yet their biological underpinnings remain inadequately understood. Here, we analyzed clinical and genomic disparities between CYAs and OAs in various hematologic malignancies. We found substantial differences in clinical features such as patient sex, ethnicity, metastasis rates, and tumor subtypes. Genomically, most CYA hematologic malignancies indicated lower mutational burden. Subsequently, we identified differentially mutated genes (DMGs) with varying mutation rates between CYAs and OAs, noting fewer mutations in CYAs for most genes such as TP53, TET2, and DNMT3A. In contrast, several DMGs (i.e., NRAS, KRAS, SMARCA4, ID3, PTPN11, WT1, and KIT) were overrepresented in CYAs. We further investigated human protein interacting partners of these identified DMGs that were highly mutated in CYAs/OAs, respectively, and found significant differences in network topological and functional roles. Notably, CYA malignancies demonstrated extensive copy number alterations (CNAs) and more driver gene fusions. In particular, four CNA differential genes (i.e., ARID1B, MYB, TP53, and ESR1) were overrepresented as amplifications and deletions in CYAs and OAs, respectively. Ultimately, we demonstrated a landscape comparative view of clinically actionable genetic events in CYAs and OAs, providing clues for age-related personalized treatment.
- Open Access
- 18 June 2024
- Published by American Society for Microbiology in mSystems
- Vol. 9 (6) , e0138523
Comparative analysis of dynamic transcriptomes reveals specific COVID-19 features and pathogenesis of immunocompromised populations
A dysfunction of human host genes and proteins in coronavirus infectious disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a key factor impacting clinical symptoms and outcomes. Yet, a detailed understanding of human host immune responses is still incomplete. Here, we applied RNA sequencing to 94 samples of COVID-19 patients with and without hematological tumors as well as COVID-19 uninfected non-tumor individuals to obtain a comprehensive transcriptome landscape of both hematological tumor patients and non-tumor individuals. In our analysis, we further accounted for the human-SARS-CoV-2 protein interactome, human protein interactome, and human protein complex subnetworks to understand the mechanisms of SARS-CoV-2 infection and host immune responses. Our data sets enabled us to identify important SARS-CoV-2 (non-)targeted differentially expressed genes and complexes post-SARS-CoV-2 infection in both hematological tumor and non-tumor individuals. We found several unique differentially expressed genes, complexes, and functions/pathways such as blood coagulation (APOE, SERPINE1, SERPINE2, and TFPI), lipoprotein particle remodeling (APOC2, APOE, and CETP), and pro-B cell differentiation (IGHM, VPREB1, and IGLL1) during COVID-19 infection in patients with hematological tumors. In particular, APOE, a gene that is associated with both blood coagulation and lipoprotein particle remodeling, is not only upregulated in hematological tumor patients post-SARS-CoV-2 infection but also significantly expressed in acute dead patients with hematological tumors, providing clues for the design of future therapeutic strategies specifically targeting COVID-19 in patients with hematological tumors. Our data provide a rich resource for understanding the specific pathogenesis of COVID-19 in immunocompromised patients, such as those with hematological malignancies, and developing effective therapeutics for COVID-19. IMPORTANCE: A majority of previous studies focused on the characterization of coronavirus infectious disease 2019 (COVID-19) disease severity in people with normal immunity, while the characterization of COVID-19 in immunocompromised populations is still limited. Our study profiles changes in the transcriptome landscape post-severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in hematological tumor patients and non-tumor individuals. Furthermore, our integrative and comparative systems biology analysis of the interactome, complexome, and transcriptome provides new insights into the tumor-specific pathogenesis of COVID-19. Our findings confirm that SARS-CoV-2 potentially tends to target more non-functional host proteins to indirectly affect host immune responses in hematological tumor patients. The identified unique genes, complexes, functions/pathways, and expression patterns post-SARS-CoV-2 infection in patients with hematological tumors increase our understanding of how SARS-CoV-2 manipulates the host molecular mechanism. Our observed differential genes/complexes and clinical indicators of normal/long infection and deceased COVID-19 patients provide clues for understanding the mechanism of COVID-19 progression in hematological tumors. Finally, our study provides an important data resource that supports the increasing value of the application of publicly accessible data sets to public health.
- 18 June 2024
- Published by American Society for Microbiology in mSystems
- Vol. 9 (6) , e0138523
Open AccessComparative analysis of dynamic transcriptomes reveals specific COVID-19 features and pathogenesis of immunocompromised populations
A dysfunction of human host genes and proteins in coronavirus infectious disease 2019 (COVID-19) caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a key factor impacting clinical symptoms and outcomes. Yet, a detailed understanding of human host immune responses is still incomplete. Here, we applied RNA sequencing to 94 samples of COVID-19 patients with and without hematological tumors as well as COVID-19 uninfected non-tumor individuals to obtain a comprehensive transcriptome landscape of both hematological tumor patients and non-tumor individuals. In our analysis, we further accounted for the human-SARS-CoV-2 protein interactome, human protein interactome, and human protein complex subnetworks to understand the mechanisms of SARS-CoV-2 infection and host immune responses. Our data sets enabled us to identify important SARS-CoV-2 (non-)targeted differentially expressed genes and complexes post-SARS-CoV-2 infection in both hematological tumor and non-tumor individuals. We found several unique differentially expressed genes, complexes, and functions/pathways such as blood coagulation (APOE, SERPINE1, SERPINE2, and TFPI), lipoprotein particle remodeling (APOC2, APOE, and CETP), and pro-B cell differentiation (IGHM, VPREB1, and IGLL1) during COVID-19 infection in patients with hematological tumors. In particular, APOE, a gene that is associated with both blood coagulation and lipoprotein particle remodeling, is not only upregulated in hematological tumor patients post-SARS-CoV-2 infection but also significantly expressed in acute dead patients with hematological tumors, providing clues for the design of future therapeutic strategies specifically targeting COVID-19 in patients with hematological tumors. Our data provide a rich resource for understanding the specific pathogenesis of COVID-19 in immunocompromised patients, such as those with hematological malignancies, and developing effective therapeutics for COVID-19. IMPORTANCE: A majority of previous studies focused on the characterization of coronavirus infectious disease 2019 (COVID-19) disease severity in people with normal immunity, while the characterization of COVID-19 in immunocompromised populations is still limited. Our study profiles changes in the transcriptome landscape post-severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in hematological tumor patients and non-tumor individuals. Furthermore, our integrative and comparative systems biology analysis of the interactome, complexome, and transcriptome provides new insights into the tumor-specific pathogenesis of COVID-19. Our findings confirm that SARS-CoV-2 potentially tends to target more non-functional host proteins to indirectly affect host immune responses in hematological tumor patients. The identified unique genes, complexes, functions/pathways, and expression patterns post-SARS-CoV-2 infection in patients with hematological tumors increase our understanding of how SARS-CoV-2 manipulates the host molecular mechanism. Our observed differential genes/complexes and clinical indicators of normal/long infection and deceased COVID-19 patients provide clues for understanding the mechanism of COVID-19 progression in hematological tumors. Finally, our study provides an important data resource that supports the increasing value of the application of publicly accessible data sets to public health.
- Open Access
- 22 January 2024
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 25 (2)
Multi-modal features-based human-herpesvirus protein–protein interaction prediction by using LightGBM
The identification of human-herpesvirus protein–protein interactions (PPIs) is an essential and important entry point to understand the mechanisms of viral infection, especially in malignant tumor patients with common herpesvirus infection. While natural language processing (NLP)-based embedding techniques have emerged as powerful approaches, the application of multi-modal embedding feature fusion to predict human-herpesvirus PPIs is still limited. Here, we established a multi-modal embedding feature fusion-based LightGBM method to predict human-herpesvirus PPIs. In particular, we applied document and graph embedding approaches to represent sequence, network and function modal features of human and herpesviral proteins. Training our LightGBM models through our compiled non-rigorous and rigorous benchmarking datasets, we obtained significantly better performance compared to individual-modal features. Furthermore, our model outperformed traditional feature encodings-based machine learning methods and state-of-the-art deep learning-based methods using various benchmarking datasets. In a transfer learning step, we show that our model that was trained on human-herpesvirus PPI dataset without cytomegalovirus data can reliably predict human-cytomegalovirus PPIs, indicating that our method can comprehensively capture multi-modal fusion features of protein interactions across various herpesvirus subtypes. The implementation of our method is available at https://github.com/XiaodiYangpku/MultimodalPPI/.
- 22 January 2024
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 25 (2)
Open AccessMulti-modal features-based human-herpesvirus protein–protein interaction prediction by using LightGBM
The identification of human-herpesvirus protein–protein interactions (PPIs) is an essential and important entry point to understand the mechanisms of viral infection, especially in malignant tumor patients with common herpesvirus infection. While natural language processing (NLP)-based embedding techniques have emerged as powerful approaches, the application of multi-modal embedding feature fusion to predict human-herpesvirus PPIs is still limited. Here, we established a multi-modal embedding feature fusion-based LightGBM method to predict human-herpesvirus PPIs. In particular, we applied document and graph embedding approaches to represent sequence, network and function modal features of human and herpesviral proteins. Training our LightGBM models through our compiled non-rigorous and rigorous benchmarking datasets, we obtained significantly better performance compared to individual-modal features. Furthermore, our model outperformed traditional feature encodings-based machine learning methods and state-of-the-art deep learning-based methods using various benchmarking datasets. In a transfer learning step, we show that our model that was trained on human-herpesvirus PPI dataset without cytomegalovirus data can reliably predict human-cytomegalovirus PPIs, indicating that our method can comprehensively capture multi-modal fusion features of protein interactions across various herpesvirus subtypes. The implementation of our method is available at https://github.com/XiaodiYangpku/MultimodalPPI/.
- Open Access
- 22 June 2023
- Published by Springer Nature in BMC Medical Education
- Vol. 23 (1) , 1-8
Application of the hybrid BOPPPS teaching model in clinical internships in gynecology
Background: The traditional gynecological teaching model is not conducive to the cultivation of trainee doctors’ clinical skills, thinking patterns and doctor‒patient communication ability. This study aims to explore the effect of the application of the hybrid BOPPPS (bridge-in, objective, preassessment, participant learning, postassessment, summary) teaching model in clinical internships in gynecology. Methods: This observational study was conducted among final-year undergraduate medical trainee doctors at Jiaxing Maternity and Child Health Care Hospital from September 2020 to June 2022. Members of the control group were introduced to the traditional teaching model, while members of the experimental group were introduced to the hybrid BOPPPS teaching model. Trainee doctors’ final examination scores and teaching satisfaction were compared. Results: The control group consisted of 114 students who entered the university to pursue undergraduate degrees in 2017, and the experimental group consisted of 121 students who entered the university to pursue undergraduate degrees in 2018. The final examination scores attained by trainee doctors in the experimental group were higher than those attained by trainee doctors in the control group (P < 0.05). The final theoretical exam scores attained by members of the control group were significantly higher than their preassessment scores (P < 0.01). The scores differed significantly between female and male subjects before the internship (p0.05). In total, 93.4% of trainee doctors in the experimental group thought that the hybrid BOPPPS teaching model helped them improve their case analysis ability, and the difference in this measure between the experimental and control groups was statistically significant (P < 0.05). A total of 89.3% of trainee doctors in the experimental group supported the promotion and application of the hybrid BOPPPS model in practice in other disciplines. Conclusion: The hybrid BOPPPS teaching model helps improve trainee doctors’ learning environment, stimulate their interest and initiative in learning, enhance their clinical practice ability and increase their satisfaction; therefore, this model is worth promoting and applying in practice in other disciplines.
- 22 June 2023
- Published by Springer Nature in BMC Medical Education
- Vol. 23 (1) , 1-8
Open AccessApplication of the hybrid BOPPPS teaching model in clinical internships in gynecology
Background: The traditional gynecological teaching model is not conducive to the cultivation of trainee doctors’ clinical skills, thinking patterns and doctor‒patient communication ability. This study aims to explore the effect of the application of the hybrid BOPPPS (bridge-in, objective, preassessment, participant learning, postassessment, summary) teaching model in clinical internships in gynecology. Methods: This observational study was conducted among final-year undergraduate medical trainee doctors at Jiaxing Maternity and Child Health Care Hospital from September 2020 to June 2022. Members of the control group were introduced to the traditional teaching model, while members of the experimental group were introduced to the hybrid BOPPPS teaching model. Trainee doctors’ final examination scores and teaching satisfaction were compared. Results: The control group consisted of 114 students who entered the university to pursue undergraduate degrees in 2017, and the experimental group consisted of 121 students who entered the university to pursue undergraduate degrees in 2018. The final examination scores attained by trainee doctors in the experimental group were higher than those attained by trainee doctors in the control group (P < 0.05). The final theoretical exam scores attained by members of the control group were significantly higher than their preassessment scores (P < 0.01). The scores differed significantly between female and male subjects before the internship (p0.05). In total, 93.4% of trainee doctors in the experimental group thought that the hybrid BOPPPS teaching model helped them improve their case analysis ability, and the difference in this measure between the experimental and control groups was statistically significant (P < 0.05). A total of 89.3% of trainee doctors in the experimental group supported the promotion and application of the hybrid BOPPPS model in practice in other disciplines. Conclusion: The hybrid BOPPPS teaching model helps improve trainee doctors’ learning environment, stimulate their interest and initiative in learning, enhance their clinical practice ability and increase their satisfaction; therefore, this model is worth promoting and applying in practice in other disciplines.
- Open Access
- 17 July 2021
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 37 (24) , 4771-4778
Transfer learning via multi-scale convolutional neural layers for human–virus protein–protein interaction prediction
To complement experimental efforts, machine learning-based computational methods are playing an increasingly important role to predict human–virus protein–protein interactions (PPIs). Furthermore, transfer learning can effectively apply prior knowledge obtained from a large source dataset/task to a small target dataset/task, improving prediction performance. To predict interactions between human and viral proteins, we combine evolutionary sequence profile features with a Siamese convolutional neural network (CNN) architecture and a multi-layer perceptron. Our architecture outperforms various feature encodings-based machine learning and state-of-the-art prediction methods. As our main contribution, we introduce two transfer learning methods (i.e. ‘frozen’ type and ‘fine-tuning’ type) that reliably predict interactions in a target human–virus domain based on training in a source human–virus domain, by retraining CNN layers. Finally, we utilize the ‘frozen’ type transfer learning approach to predict human–SARS-CoV-2 PPIs, indicating that our predictions are topologically and functionally similar to experimentally known interactions. The source codes and datasets are available at https://github.com/XiaodiYangCAU/TransPPI/. Supplementary data are available at Bioinformatics online.
- 17 July 2021
- Published by Oxford University Press (OUP) in Bioinformatics
- Vol. 37 (24) , 4771-4778
Open AccessTransfer learning via multi-scale convolutional neural layers for human–virus protein–protein interaction prediction
To complement experimental efforts, machine learning-based computational methods are playing an increasingly important role to predict human–virus protein–protein interactions (PPIs). Furthermore, transfer learning can effectively apply prior knowledge obtained from a large source dataset/task to a small target dataset/task, improving prediction performance. To predict interactions between human and viral proteins, we combine evolutionary sequence profile features with a Siamese convolutional neural network (CNN) architecture and a multi-layer perceptron. Our architecture outperforms various feature encodings-based machine learning and state-of-the-art prediction methods. As our main contribution, we introduce two transfer learning methods (i.e. ‘frozen’ type and ‘fine-tuning’ type) that reliably predict interactions in a target human–virus domain based on training in a source human–virus domain, by retraining CNN layers. Finally, we utilize the ‘frozen’ type transfer learning approach to predict human–SARS-CoV-2 PPIs, indicating that our predictions are topologically and functionally similar to experimentally known interactions. The source codes and datasets are available at https://github.com/XiaodiYangCAU/TransPPI/. Supplementary data are available at Bioinformatics online.
- 8 March 2021
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 22 (5)
Current status and future perspectives of computational studies on human–virus protein–protein interactions
The protein–protein interactions (PPIs) between human and viruses mediate viral infection and host immunity processes. Therefore, the study of human–virus PPIs can help us understand the principles of human–virus relationships and can thus guide the development of highly effective drugs to break the transmission of viral infectious diseases. Recent years have witnessed the rapid accumulation of experimentally identified human–virus PPI data, which provides an unprecedented opportunity for bioinformatics studies revolving around human–virus PPIs. In this article, we provide a comprehensive overview of computational studies on human–virus PPIs, especially focusing on the method development for human–virus PPI predictions. We briefly introduce the experimental detection methods and existing database resources of human–virus PPIs, and then discuss the research progress in the development of computational prediction methods. In particular, we elaborate the machine learning-based prediction methods and highlight the need to embrace state-of-the-art deep-learning algorithms and new feature engineering techniques (e.g. the protein embedding technique derived from natural language processing). To further advance the understanding in this research topic, we also outline the practical applications of the human–virus interactome in fundamental biological discovery and new antiviral therapy development.
- 8 March 2021
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 22 (5)
Current status and future perspectives of computational studies on human–virus protein–protein interactions
The protein–protein interactions (PPIs) between human and viruses mediate viral infection and host immunity processes. Therefore, the study of human–virus PPIs can help us understand the principles of human–virus relationships and can thus guide the development of highly effective drugs to break the transmission of viral infectious diseases. Recent years have witnessed the rapid accumulation of experimentally identified human–virus PPI data, which provides an unprecedented opportunity for bioinformatics studies revolving around human–virus PPIs. In this article, we provide a comprehensive overview of computational studies on human–virus PPIs, especially focusing on the method development for human–virus PPI predictions. We briefly introduce the experimental detection methods and existing database resources of human–virus PPIs, and then discuss the research progress in the development of computational prediction methods. In particular, we elaborate the machine learning-based prediction methods and highlight the need to embrace state-of-the-art deep-learning algorithms and new feature engineering techniques (e.g. the protein embedding technique derived from natural language processing). To further advance the understanding in this research topic, we also outline the practical applications of the human–virus interactome in fundamental biological discovery and new antiviral therapy development.
- 30 January 2021
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 22 (2) , 832-844
HVIDB: a comprehensive database for human–virus protein–protein interactions
While leading to millions of people’s deaths every year the treatment of viral infectious diseases remains a huge public health challenge.Therefore, an in-depth understanding of human–virus protein–protein interactions (PPIs) as the molecular interface between a virus and its host cell is of paramount importance to obtain new insights into the pathogenesis of viral infections and development of antiviral therapeutic treatments. However, current human–virus PPI database resources are incomplete, lack annotation and usually do not provide the opportunity to computationally predict human–virus PPIs. Here, we present the Human–Virus Interaction DataBase (HVIDB, http://zzdlab.com/hvidb/) that provides comprehensively annotated human–virus PPI data as well as seamlessly integrates online PPI prediction tools. Currently, HVIDB highlights 48 643 experimentally verified human–virus PPIs covering 35 virus families, 6633 virally targeted host complexes, 3572 host dependency/restriction factors as well as 911 experimentally verified/predicted 3D complex structures of human–virus PPIs. Furthermore, our database resource provides tissue-specific expression profiles of 6790 human genes that are targeted by viruses and 129 Gene Expression Omnibus series of differentially expressed genes post-viral infections. Based on these multifaceted and annotated data, our database allows the users to easily obtain reliable information about PPIs of various human viruses and conduct an in-depth analysis of their inherent biological significance. In particular, HVIDB also integrates well-performing machine learning models to predict interactions between the human host and viral proteins that are based on (i) sequence embedding techniques, (ii) interolog mapping and (iii) domain–domain interaction inference. We anticipate that HVIDB will serve as a one-stop knowledge base to further guide hypothesis-driven experimental efforts to investigate human–virus relationships.
- 30 January 2021
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 22 (2) , 832-844
HVIDB: a comprehensive database for human–virus protein–protein interactions
While leading to millions of people’s deaths every year the treatment of viral infectious diseases remains a huge public health challenge.Therefore, an in-depth understanding of human–virus protein–protein interactions (PPIs) as the molecular interface between a virus and its host cell is of paramount importance to obtain new insights into the pathogenesis of viral infections and development of antiviral therapeutic treatments. However, current human–virus PPI database resources are incomplete, lack annotation and usually do not provide the opportunity to computationally predict human–virus PPIs. Here, we present the Human–Virus Interaction DataBase (HVIDB, http://zzdlab.com/hvidb/) that provides comprehensively annotated human–virus PPI data as well as seamlessly integrates online PPI prediction tools. Currently, HVIDB highlights 48 643 experimentally verified human–virus PPIs covering 35 virus families, 6633 virally targeted host complexes, 3572 host dependency/restriction factors as well as 911 experimentally verified/predicted 3D complex structures of human–virus PPIs. Furthermore, our database resource provides tissue-specific expression profiles of 6790 human genes that are targeted by viruses and 129 Gene Expression Omnibus series of differentially expressed genes post-viral infections. Based on these multifaceted and annotated data, our database allows the users to easily obtain reliable information about PPIs of various human viruses and conduct an in-depth analysis of their inherent biological significance. In particular, HVIDB also integrates well-performing machine learning models to predict interactions between the human host and viral proteins that are based on (i) sequence embedding techniques, (ii) interolog mapping and (iii) domain–domain interaction inference. We anticipate that HVIDB will serve as a one-stop knowledge base to further guide hypothesis-driven experimental efforts to investigate human–virus relationships.