Parvin Razzaghi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Parvin Razzaghi has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Karim Abbasi
Mosaheb Institute for Mathematical Research, Kharazmi University
6 shared publications - Sajjad Gharaghani
Laboratory of Bioinformatics and Drug Design (LBD), Institute of Biochemistry and Biophysics, University of Tehran, Tehran, Iran
4 shared publications - Ali Masoudi-Nejad
Laboratory of Systems Biology and Bioinformatics (LBB), Department of Bioinformatics, Kish International Campus, University of Tehran, Kish Island, Iran
2 shared publications - Alireza Dehghan
Department of Computer Engineering, Faculty of Technology and Engineering, Salman Farsi University of Kazerun, Kazerun, Iran
2 shared publications - Ebrahim Ansari
Institute for Advanced Studies in Basic Siences (IASBS)
2 shared publications Hadi Abdi Khojasteh
Institute for Advanced Studies in Basic Sciences (IASBS) Zanjan Iran
2 shared publications
Latest Publications
- Open Access
- 18 May 2026
- Vol. 6  (1)
CDCR-Rank: a computational model for predicting drug combination dose response using ranking-based optimization
Predicting synergistic drug combinations for cancer therapies is vital but challenged by interaction complexity and vast combinatorial possibilities. To address this, we propose a novel ranking-based approach for drug synergy prediction. Our method builds upon the pre-trained CDCR model to predict absolute synergy scores and introduces a novel architecture to directly compare a list of combinations. A key innovation is our custom loss function, which combines mean squared error with a ranking loss term to simultaneously learn accurate synergy values and their correct relative ordering. We comprehensively evaluated our model, CDCR-Rank, on the NCI-ALMANAC benchmark under three challenging scenarios, including predicting synergy for novel drug pairs without monotherapy data. CDCR-Rank demonstrated superior performance, consistently outperforming state-of-the-art methods such as comboFM and comboLTR. Ablation studies validated the importance of each component: a one-dimensional convolutional neural network (1D-CNN) for generating task-specific drug representations from SMILES strings, a sinusoidal encoder for modeling non-linear dose-response curves, and the uRank loss for optimizing combination rankings. Our integrated framework achieves significantly improved accuracy and robustness in predicting drug synergy. By reliably prioritizing the most promising combinations for experimental validation, this work has the potential to accelerate the discovery of effective multi-drug cancer therapies. CDCR-Rank is implemented in Python and is available at https://github.com/ParvinRazzaghi/CDCR-Rank
- 18 May 2026
- Vol. 6  (1)
Open AccessCDCR-Rank: a computational model for predicting drug combination dose response using ranking-based optimization
Predicting synergistic drug combinations for cancer therapies is vital but challenged by interaction complexity and vast combinatorial possibilities. To address this, we propose a novel ranking-based approach for drug synergy prediction. Our method builds upon the pre-trained CDCR model to predict absolute synergy scores and introduces a novel architecture to directly compare a list of combinations. A key innovation is our custom loss function, which combines mean squared error with a ranking loss term to simultaneously learn accurate synergy values and their correct relative ordering. We comprehensively evaluated our model, CDCR-Rank, on the NCI-ALMANAC benchmark under three challenging scenarios, including predicting synergy for novel drug pairs without monotherapy data. CDCR-Rank demonstrated superior performance, consistently outperforming state-of-the-art methods such as comboFM and comboLTR. Ablation studies validated the importance of each component: a one-dimensional convolutional neural network (1D-CNN) for generating task-specific drug representations from SMILES strings, a sinusoidal encoder for modeling non-linear dose-response curves, and the uRank loss for optimizing combination rankings. Our integrated framework achieves significantly improved accuracy and robustness in predicting drug synergy. By reliably prioritizing the most promising combinations for experimental validation, this work has the potential to accelerate the discovery of effective multi-drug cancer therapies. CDCR-Rank is implemented in Python and is available at https://github.com/ParvinRazzaghi/CDCR-Rank
- Open Access
- 8 May 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21  (5) , e0348908
DeepDRP: Dose-response predictions of drug pairs using deep learning based on data-driven feature representation and dose-response curve characteristics
Combination therapies have become a cornerstone of modern medicine, offering improved treatment outcomes and reduced side effects compared to monotherapies. However, the efficacy and safety of drug combinations depend heavily on the specific doses of each component, making the optimization of dosing regimens a crucial yet challenging task. Despite the importance of dose optimization, few computational methods address this challenge. Here, we propose DeepDRP, a novel approach that integrates two complementary models: the global model, which is trained using all available combinations, and the local model, which utilizes only samples with similar information to queries. The DeepDRP architecture comprises three main modules: the first module has three embedding networks to extract features from drugs, doses, and cell lines, and then predicts synergy using the fused knowledge. The second module constructs a graph using the input query and the set of similar retrieved training samples fed into a semi-supervised graph convolutional network to predict the synergy value. Finally, these two models are aggregated to have the final predicted value. To evaluate the proposed approach, it is applied to the NCI-ALMANAC dataset and O’Neil dataset. The obtained results denote that the proposed method achieves superior performance with respect to the other approaches.
- 8 May 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21  (5) , e0348908
Open AccessDeepDRP: Dose-response predictions of drug pairs using deep learning based on data-driven feature representation and dose-response curve characteristics
Combination therapies have become a cornerstone of modern medicine, offering improved treatment outcomes and reduced side effects compared to monotherapies. However, the efficacy and safety of drug combinations depend heavily on the specific doses of each component, making the optimization of dosing regimens a crucial yet challenging task. Despite the importance of dose optimization, few computational methods address this challenge. Here, we propose DeepDRP, a novel approach that integrates two complementary models: the global model, which is trained using all available combinations, and the local model, which utilizes only samples with similar information to queries. The DeepDRP architecture comprises three main modules: the first module has three embedding networks to extract features from drugs, doses, and cell lines, and then predicts synergy using the fused knowledge. The second module constructs a graph using the input query and the set of similar retrieved training samples fed into a semi-supervised graph convolutional network to predict the synergy value. Finally, these two models are aggregated to have the final predicted value. To evaluate the proposed approach, it is applied to the NCI-ALMANAC dataset and O’Neil dataset. The obtained results denote that the proposed method achieves superior performance with respect to the other approaches.
- Open Access
- 4 May 2026
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 27  (3)
MT-ConBiFormer-GPT: multi-target molecular generation for low-data drug discovery via a contrastive BiFormer-GPT architecture and curriculum learning with cross-domain generalization
Multi-target compounds, or polypharmacological agents, hold significant potential for complex diseases like cancer, where single-target therapies are often insufficient. A lack of high-quality bioactivity data limits progress in this field, especially for compounds interacting with multiple proteins simultaneously. This study introduces MT-ConBiFormer-GPT, a deep generative model designed explicitly for low-data, multi-target molecular generation, focusing on the critical PI3K–AKT–mTOR cancer signaling pathway. The framework integrates a variational autoencoder with a BiFormer encoder to capture long-range dependencies in SMILES strings, reducing the quadratic computational complexity associated with standard transformers and mitigating semantic discontinuities. It employs a SMILES-GPT decoder for progressive molecule generation and follows a three-phase training pipeline: unsupervised pre-training, supervised contrastive learning, and curriculum-based fine-tuning. The framework’s efficacy was evaluated through a rigorous, multi-stage assessment. First, the framework was evaluated through benchmarking against state-of-the-art models, with a specialized head-to-head variant, MT-ConBiFormer-GPT_H2H, demonstrating superior performance, thereby validating its generalizability from oncology to neuropsychiatry. An internal ablation study further revealed that the full MT-ConBiFormer-GPT significantly outperformed its baseline, MT-BiFormer-GPT, in both dual- and triplet-target generation tasks, highlighting the advantages of the contrastive learning stage. Additionally, the foundational Base-BiFormer-GPT architecture, a model lacking both the contrastive and curriculum learning stages, highlighted its intrinsic robustness by achieving competitive outcomes in a distinct omics-driven design task. Docking simulations and mechanistic analyses show that the generated molecules, including high-fidelity and scaffold-hopping candidates, display more favorable binding modes than reference inhibitors. This study presents a flexible and computationally efficient framework for multi-target drug discovery in data-limited settings.
- 4 May 2026
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 27  (3)
Open AccessMT-ConBiFormer-GPT: multi-target molecular generation for low-data drug discovery via a contrastive BiFormer-GPT architecture and curriculum learning with cross-domain generalization
Multi-target compounds, or polypharmacological agents, hold significant potential for complex diseases like cancer, where single-target therapies are often insufficient. A lack of high-quality bioactivity data limits progress in this field, especially for compounds interacting with multiple proteins simultaneously. This study introduces MT-ConBiFormer-GPT, a deep generative model designed explicitly for low-data, multi-target molecular generation, focusing on the critical PI3K–AKT–mTOR cancer signaling pathway. The framework integrates a variational autoencoder with a BiFormer encoder to capture long-range dependencies in SMILES strings, reducing the quadratic computational complexity associated with standard transformers and mitigating semantic discontinuities. It employs a SMILES-GPT decoder for progressive molecule generation and follows a three-phase training pipeline: unsupervised pre-training, supervised contrastive learning, and curriculum-based fine-tuning. The framework’s efficacy was evaluated through a rigorous, multi-stage assessment. First, the framework was evaluated through benchmarking against state-of-the-art models, with a specialized head-to-head variant, MT-ConBiFormer-GPT_H2H, demonstrating superior performance, thereby validating its generalizability from oncology to neuropsychiatry. An internal ablation study further revealed that the full MT-ConBiFormer-GPT significantly outperformed its baseline, MT-BiFormer-GPT, in both dual- and triplet-target generation tasks, highlighting the advantages of the contrastive learning stage. Additionally, the foundational Base-BiFormer-GPT architecture, a model lacking both the contrastive and curriculum learning stages, highlighted its intrinsic robustness by achieving competitive outcomes in a distinct omics-driven design task. Docking simulations and mechanistic analyses show that the generated molecules, including high-fidelity and scaffold-hopping candidates, display more favorable binding modes than reference inhibitors. This study presents a flexible and computationally efficient framework for multi-target drug discovery in data-limited settings.
- Open Access
Detection of Pediatric Dental Caries in Panoramic Radiograph Using Deep Learning: A Benchmark Study on MD-OPG
Early detection of dental caries in children is critical to prevent irreversible tooth damage and guarantee optimal oral health outcomes. However, interpreting pediatric panoramic radiographs throughout the mixed dentition stage remains a very challenging task due to overlap in anatomical structures and developmental variability. This complexity underscores the need for well curated, representative datasets that enable the development of reliable computer-aided diagnostic models. Herein, this study introduces the Mixed Dentition Orthopantomogram Dataset, a newly developed, publicly available dataset of children that was carefully labeled by dental specialists to identify proximal and occlusal caries regions in the range of 3–12 years. To evaluate the dataset’s applicability for artificial intelligence research, we benchmarked it using both classification and segmentation models. A patch-based classifier achieved an average AUC of 0.89 and Recall 0.85 in distinguishing healthy and carious regions. For segmentation, we evaluated U-Net and Attention U-Net with multiple loss functions, and the Attention U-Net trained with Focal loss achieved the best Dice score of 0.94. Collectively, these findings support the dataset’s utility for pediatric caries analysis and demonstrate the viability of deep learning approaches for mixed dentition panoramic imaging.
Open AccessDetection of Pediatric Dental Caries in Panoramic Radiograph Using Deep Learning: A Benchmark Study on MD-OPG
Early detection of dental caries in children is critical to prevent irreversible tooth damage and guarantee optimal oral health outcomes. However, interpreting pediatric panoramic radiographs throughout the mixed dentition stage remains a very challenging task due to overlap in anatomical structures and developmental variability. This complexity underscores the need for well curated, representative datasets that enable the development of reliable computer-aided diagnostic models. Herein, this study introduces the Mixed Dentition Orthopantomogram Dataset, a newly developed, publicly available dataset of children that was carefully labeled by dental specialists to identify proximal and occlusal caries regions in the range of 3–12 years. To evaluate the dataset’s applicability for artificial intelligence research, we benchmarked it using both classification and segmentation models. A patch-based classifier achieved an average AUC of 0.89 and Recall 0.85 in distinguishing healthy and carious regions. For segmentation, we evaluated U-Net and Attention U-Net with multiple loss functions, and the Attention U-Net trained with Focal loss achieved the best Dice score of 0.94. Collectively, these findings support the dataset’s utility for pediatric caries analysis and demonstrate the viability of deep learning approaches for mixed dentition panoramic imaging.
- 3 June 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 194, 110442
DFT_ANPD: A dual-feature two-sided attention network for anticancer natural products detection
- 3 June 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 194, 110442
DFT_ANPD: A dual-feature two-sided attention network for anticancer natural products detection
- Open Access
- 28 February 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15  (1) , 1-13
Integrating convolutional layers and biformer network with forward-forward and backpropagation training
Accurate molecular property prediction is crucial for drug discovery and computational chemistry, facilitating the identification of promising compounds and accelerating therapeutic development. Traditional machine learning falters with high-dimensional data and manual feature engineering, while existing deep learning approaches may not capture complex molecular structures, leaving a research gap. We introduce Deep-CBN, a novel framework designed to enhance molecular property prediction by capturing intricate molecular representations directly from raw data, thus improving accuracy and efficiency. Our methodology combines convolutional neural networks (CNNs) with a BiFormer attention mechanism, employing both the forward-forward algorithm and backpropagation. The model operates in three stages: (1) feature learning, extracting local features from SMILES strings using CNNs; (2) attention refinement, capturing global context with a BiFormer module enhanced by the forward-forward algorithm; and (3) prediction subnetwork tuning, fine-tuning via backpropagation. Evaluations on benchmark datasets—including Tox21, BBBP, SIDER, ClinTox, BACE, HIV, and MUV—show that Deep-CBN achieves near-perfect ROC-AUC scores, significantly outperforming state-of-the-art methods. These findings demonstrate its effectiveness in capturing complex molecular patterns, offering a robust tool to accelerate drug discovery processes.
- 28 February 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15  (1) , 1-13
Open AccessIntegrating convolutional layers and biformer network with forward-forward and backpropagation training
Accurate molecular property prediction is crucial for drug discovery and computational chemistry, facilitating the identification of promising compounds and accelerating therapeutic development. Traditional machine learning falters with high-dimensional data and manual feature engineering, while existing deep learning approaches may not capture complex molecular structures, leaving a research gap. We introduce Deep-CBN, a novel framework designed to enhance molecular property prediction by capturing intricate molecular representations directly from raw data, thus improving accuracy and efficiency. Our methodology combines convolutional neural networks (CNNs) with a BiFormer attention mechanism, employing both the forward-forward algorithm and backpropagation. The model operates in three stages: (1) feature learning, extracting local features from SMILES strings using CNNs; (2) attention refinement, capturing global context with a BiFormer module enhanced by the forward-forward algorithm; and (3) prediction subnetwork tuning, fine-tuning via backpropagation. Evaluations on benchmark datasets—including Tox21, BBBP, SIDER, ClinTox, BACE, HIV, and MUV—show that Deep-CBN achieves near-perfect ROC-AUC scores, significantly outperforming state-of-the-art methods. These findings demonstrate its effectiveness in capturing complex molecular patterns, offering a robust tool to accelerate drug discovery processes.
- Open Access
- 18 November 2024
- Published by SCI AND TECH UNIVERSAL INC in Scientia Iranica
Perceptual Deep Portrait Image Selection: Subjective and Objective Approaches
The efficiency of portrait image selection and analysis systems is completely dependent on the quality of the face image, which depends on various factors. Since real-time manual selection of high-quality portrait photos from a sequence of different frames or images is usually impossible, using automatic methods can be useful in selecting photos, especially in large collections. Existing automatic methods may not be able to perform like humans in portrait classification. These methods may consider only special factors like emotional state or gaze direction to select an image. In this work, a large collection of facial images was collected, and under a subjective quality assessment study, they were judged by more than 80 people. A deep classifier network using transfer learning and the fine-tuning approach is proposed, which is learned end-to-end having the subjective labels to select good portrait images objectively. Quantitative and qualitative results show that this model performs better than state-of-the-art image classification networks. In addition, our qualitative evaluation showed that our model can separate good portrait images in the way that humans do. Therefore, this model can be reliably used in mobile phones, digital Cameras, and other imaging systems.
- 18 November 2024
- Published by SCI AND TECH UNIVERSAL INC in Scientia Iranica
Open AccessPerceptual Deep Portrait Image Selection: Subjective and Objective Approaches
The efficiency of portrait image selection and analysis systems is completely dependent on the quality of the face image, which depends on various factors. Since real-time manual selection of high-quality portrait photos from a sequence of different frames or images is usually impossible, using automatic methods can be useful in selecting photos, especially in large collections. Existing automatic methods may not be able to perform like humans in portrait classification. These methods may consider only special factors like emotional state or gaze direction to select an image. In this work, a large collection of facial images was collected, and under a subjective quality assessment study, they were judged by more than 80 people. A deep classifier network using transfer learning and the fine-tuning approach is proposed, which is learned end-to-end having the subjective labels to select good portrait images objectively. Quantitative and qualitative results show that this model performs better than state-of-the-art image classification networks. In addition, our qualitative evaluation showed that our model can separate good portrait images in the way that humans do. Therefore, this model can be reliably used in mobile phones, digital Cameras, and other imaging systems.
- Open Access
- 24 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 115-121
ProtContext-DTI: Protein Contextual Representation Using Masked Language Modeling in Drug Target Interaction Prediction
- 24 May 2024
- Published by Association for Computing Machinery (ACM)
- p. 115-121
Open AccessProtContext-DTI: Protein Contextual Representation Using Masked Language Modeling in Drug Target Interaction Prediction
- 22 March 2024
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 64  (7) , 2577-2585
CFSSynergy: Combining Feature-Based and Similarity-Based Methods for Drug Synergy Prediction
Drug synergy prediction plays a vital role in cancer treatment. Because experimental approaches are labor-intensive and expensive, computational-based approaches get more attention. There are two types of computational methods for drug synergy prediction: feature-based and similarity-based. In feature-based methods, the main focus is to extract more discriminative features from drug pairs and cell lines to pass to the task predictor. In similarity-based methods, the similarities among all drugs and cell lines are utilized as features and fed into the task predictor. In this work, a novel approach, called CFSSynergy, that combines these two viewpoints is proposed. First, a discriminative representation is extracted for paired drugs and cell lines as input. We have utilized transformer-based architecture for drugs. For cell lines, we have created a similarity matrix between proteins using the Node2Vec algorithm. Then, the new cell line representation is computed by multiplying the protein-protein similarity matrix and the initial cell line representation. Next, we compute the similarity between unique drugs and unique cells using the learned representation for paired drugs and cell lines. Then, we compute a new representation for paired drugs and cell lines based on the similarity-based features and the learned features. Finally, these features are fed to XGBoost as a task predictor. Two well-known data sets were used to evaluate the performance of our proposed method: DrugCombDB and OncologyScreen. The CFSSynergy approach consistently outperformed existing methods in comparative evaluations. This substantiates the efficacy of our approach in capturing complex synergistic interactions between drugs and cell lines, setting it apart from conventional similarity-based or feature-based methods.
- 22 March 2024
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 64  (7) , 2577-2585
CFSSynergy: Combining Feature-Based and Similarity-Based Methods for Drug Synergy Prediction
Drug synergy prediction plays a vital role in cancer treatment. Because experimental approaches are labor-intensive and expensive, computational-based approaches get more attention. There are two types of computational methods for drug synergy prediction: feature-based and similarity-based. In feature-based methods, the main focus is to extract more discriminative features from drug pairs and cell lines to pass to the task predictor. In similarity-based methods, the similarities among all drugs and cell lines are utilized as features and fed into the task predictor. In this work, a novel approach, called CFSSynergy, that combines these two viewpoints is proposed. First, a discriminative representation is extracted for paired drugs and cell lines as input. We have utilized transformer-based architecture for drugs. For cell lines, we have created a similarity matrix between proteins using the Node2Vec algorithm. Then, the new cell line representation is computed by multiplying the protein-protein similarity matrix and the initial cell line representation. Next, we compute the similarity between unique drugs and unique cells using the learned representation for paired drugs and cell lines. Then, we compute a new representation for paired drugs and cell lines based on the similarity-based features and the learned features. Finally, these features are fed to XGBoost as a task predictor. Two well-known data sets were used to evaluate the performance of our proposed method: DrugCombDB and OncologyScreen. The CFSSynergy approach consistently outperformed existing methods in comparative evaluations. This substantiates the efficacy of our approach in capturing complex synergistic interactions between drugs and cell lines, setting it apart from conventional similarity-based or feature-based methods.
- Open Access
- 30 January 2024
- Published by Springer Nature in BMC Bioinformatics
- Vol. 25  (1) , 1-15
CCL-DTI: contributing the contrastive loss in drug–target interaction prediction
Background: The Drug–Target Interaction (DTI) prediction uses a drug molecule and a protein sequence as inputs to predict the binding affinity value. In recent years, deep learning-based models have gotten more attention. These methods have two modules: the feature extraction module and the task prediction module. In most deep learning-based approaches, a simple task prediction loss (i.e., categorical cross entropy for the classification task and mean squared error for the regression task) is used to learn the model. In machine learning, contrastive-based loss functions are developed to learn more discriminative feature space. In a deep learning-based model, extracting more discriminative feature space leads to performance improvement for the task prediction module. Results: In this paper, we have used multimodal knowledge as input and proposed an attention-based fusion technique to combine this knowledge. Also, we investigate how utilizing contrastive loss function along the task prediction loss could help the approach to learn a more powerful model. Four contrastive loss functions are considered: (1) max-margin contrastive loss function, (2) triplet loss function, (3) Multi-class N-pair Loss Objective, and (4) NT-Xent loss function. The proposed model is evaluated using four well-known datasets: Wang et al. dataset, Luo's dataset, Davis, and KIBA datasets. Conclusions: Accordingly, after reviewing the state-of-the-art methods, we developed a multimodal feature extraction network by combining protein sequences and drug molecules, along with protein–protein interaction networks and drug–drug interaction networks. The results show it performs significantly better than the comparable state-of-the-art approaches.
- 30 January 2024
- Published by Springer Nature in BMC Bioinformatics
- Vol. 25  (1) , 1-15
Open AccessCCL-DTI: contributing the contrastive loss in drug–target interaction prediction
Background: The Drug–Target Interaction (DTI) prediction uses a drug molecule and a protein sequence as inputs to predict the binding affinity value. In recent years, deep learning-based models have gotten more attention. These methods have two modules: the feature extraction module and the task prediction module. In most deep learning-based approaches, a simple task prediction loss (i.e., categorical cross entropy for the classification task and mean squared error for the regression task) is used to learn the model. In machine learning, contrastive-based loss functions are developed to learn more discriminative feature space. In a deep learning-based model, extracting more discriminative feature space leads to performance improvement for the task prediction module. Results: In this paper, we have used multimodal knowledge as input and proposed an attention-based fusion technique to combine this knowledge. Also, we investigate how utilizing contrastive loss function along the task prediction loss could help the approach to learn a more powerful model. Four contrastive loss functions are considered: (1) max-margin contrastive loss function, (2) triplet loss function, (3) Multi-class N-pair Loss Objective, and (4) NT-Xent loss function. The proposed model is evaluated using four well-known datasets: Wang et al. dataset, Luo's dataset, Davis, and KIBA datasets. Conclusions: Accordingly, after reviewing the state-of-the-art methods, we developed a multimodal feature extraction network by combining protein sequences and drug molecules, along with protein–protein interaction networks and drug–drug interaction networks. The results show it performs significantly better than the comparable state-of-the-art approaches.