Arkaprava Banerjee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Arkaprava Banerjee has a total of 59 co-authors. Here are the top co-authors by number of shared publications:
- Kunal Roy
Jadavpur University
53 shared publications - Supratik Kar
Kean University
8 shared publications - Vinay Kumar
Jadavpur University
6 shared publications Mainak Chatterjee
Drug Theoretics and Cheminformatics Laboratory, Department of Pharmaceutical Technology, Jadavpur University, Kolkata 700032, India
4 shared publicationsPriyanka De
4 shared publications- Agnieszka Gajewicz
Department of Physical Chemistry, Faculty of Chemistry, Gdansk University of Technology, Gabriela Narutowicza Str. 11/12, Gdansk 80-233, Poland
3 shared publications
Latest Publications
- 14 August 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 66 (17) , 10835-10846
Q performance 2: Toward Unbiased Selection of Machine-Learning Regression Models
Selecting a single best machine-learning regression model from a set of competing models can be challenging. While models selected based on cross-validation performance do not guarantee good predictions on external data, models selected solely on external validation performance do not ascertain precise predictions for other external sets. Therefore, we propose three quantitative metrics to guide modelers in selecting the best model using the modeling set information only. Three quantitative data sets of varying sizes and complexities were considered. Each data set was randomly split into a modeling set and an independent test set. The modeling set was further split thrice to generate training and validation sets. Various machine-learning models were developed and validated against the validation sets. Our proposed metrics were computed for each model using only training and validation set performances. The novel metric values guided the selection of the best models, which also demonstrated expected performance on the independent test set. Furthermore, successful applications of our framework on two additional benchmark data sets demonstrated its wider generalizability.
- 14 August 2026
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 66 (17) , 10835-10846
Q performance 2: Toward Unbiased Selection of Machine-Learning Regression Models
Selecting a single best machine-learning regression model from a set of competing models can be challenging. While models selected based on cross-validation performance do not guarantee good predictions on external data, models selected solely on external validation performance do not ascertain precise predictions for other external sets. Therefore, we propose three quantitative metrics to guide modelers in selecting the best model using the modeling set information only. Three quantitative data sets of varying sizes and complexities were considered. Each data set was randomly split into a modeling set and an independent test set. The modeling set was further split thrice to generate training and validation sets. Various machine-learning models were developed and validated against the validation sets. Our proposed metrics were computed for each model using only training and validation set performances. The novel metric values guided the selection of the best models, which also demonstrated expected performance on the independent test set. Furthermore, successful applications of our framework on two additional benchmark data sets demonstrated its wider generalizability.
- 8 June 2026
- Published by Springer Nature in Journal of Computer-Aided Molecular Design
- Vol. 40 (1) , 143
A round-robin exercise for the precise prediction of aqueous solubility of organic chemicals using chemometric, machine learning, and stacking ensemble of deep learning models
Aqueous solubility is an important property for assessing the druggability and ecotoxicological effects of molecules. Successful drug candidates should have optimal aqueous solubility to improve bioavailability to target tissues. To effectively screen molecules in a short period of time, reliable predictive models are highly useful. In the present study, we conducted a round-robin exercise using a large, curated dataset of over 6000 compounds to predict aqueous solubility quantitatively. The six participating groups used an array of Machine Learning and Deep Learning algorithms to develop models with strong robustness and external predictive performance. All the models underwent rigorous Leave-One-Out and tenfold cross-validation. The diversity of training sets and descriptor types used by different groups paved the way for exploring the mechanistic basis for the efficient identification of contributing features. The best-performing model was selected using the statistical Sum of Ranking Differences (SRD) approach, considering the performances on training, cross-validation, and test, as well as the performance difference between the training and test sets. Additionally, a curated, true external set was screened by the six different models. Here, the best-performing model was selected using a consensus ranking strategy based on Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and \(R_{Ext}^{2}\). In both approaches, i.e., the inherent model performance in terms of training, test, and cross-validation statistics, and the ability of the model to efficiently predict true external data, the Stacking Ensemble of Deep q-RASPR models emerged as the winner. This model showed comparable predictive performance to the previously reported model, which apparently lacked a proper data curation workflow and contained a significant number of duplicates and mixtures in its dataset, which can inflate model statistics. The insights from the different feature contributions from the different groups identified the useful structural and physicochemical aspects, which can help synthetic chemists to optimize molecules. Graphical abstract The alternative text for this image may have been generated using AI. In the present study, we conducted a round-robin exercise using a large, curated dataset of over 6000 compounds to predict aqueous solubility quantitatively- 8 June 2026
- Published by Springer Nature in Journal of Computer-Aided Molecular Design
- Vol. 40 (1) , 143
A round-robin exercise for the precise prediction of aqueous solubility of organic chemicals using chemometric, machine learning, and stacking ensemble of deep learning models
Aqueous solubility is an important property for assessing the druggability and ecotoxicological effects of molecules. Successful drug candidates should have optimal aqueous solubility to improve bioavailability to target tissues. To effectively screen molecules in a short period of time, reliable predictive models are highly useful. In the present study, we conducted a round-robin exercise using a large, curated dataset of over 6000 compounds to predict aqueous solubility quantitatively. The six participating groups used an array of Machine Learning and Deep Learning algorithms to develop models with strong robustness and external predictive performance. All the models underwent rigorous Leave-One-Out and tenfold cross-validation. The diversity of training sets and descriptor types used by different groups paved the way for exploring the mechanistic basis for the efficient identification of contributing features. The best-performing model was selected using the statistical Sum of Ranking Differences (SRD) approach, considering the performances on training, cross-validation, and test, as well as the performance difference between the training and test sets. Additionally, a curated, true external set was screened by the six different models. Here, the best-performing model was selected using a consensus ranking strategy based on Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and \(R_{Ext}^{2}\). In both approaches, i.e., the inherent model performance in terms of training, test, and cross-validation statistics, and the ability of the model to efficiently predict true external data, the Stacking Ensemble of Deep q-RASPR models emerged as the winner. This model showed comparable predictive performance to the previously reported model, which apparently lacked a proper data curation workflow and contained a significant number of duplicates and mixtures in its dataset, which can inflate model statistics. The insights from the different feature contributions from the different groups identified the useful structural and physicochemical aspects, which can help synthetic chemists to optimize molecules. Graphical abstract The alternative text for this image may have been generated using AI. In the present study, we conducted a round-robin exercise using a large, curated dataset of over 6000 compounds to predict aqueous solubility quantitatively- 12 March 2026
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 65 (11) , 5954-5970
Machine Learning-Based QSPR Modeling of log CMC Values of Per- and Polyfluoroalkyl Substances (PFASs) and the Identification of Property Cliffs
- 12 March 2026
- Published by American Chemical Society (ACS) in Industrial & Engineering Chemistry Research
- Vol. 65 (11) , 5954-5970
Machine Learning-Based QSPR Modeling of log CMC Values of Per- and Polyfluoroalkyl Substances (PFASs) and the Identification of Property Cliffs
- 8 January 2026
- Published by Royal Society of Chemistry (RSC) in Environmental Science: Processes & Impacts
- Vol. 28 (3) , 699-711
Comparative machine learning and deep learning frameworks for robust carcinogenicity prediction and activity cliff analysis
In this study, we developed predictive models for binary carcinogenicity data in rats using a range of feature-based and chemical language modeling approaches.
- 8 January 2026
- Published by Royal Society of Chemistry (RSC) in Environmental Science: Processes & Impacts
- Vol. 28 (3) , 699-711
Comparative machine learning and deep learning frameworks for robust carcinogenicity prediction and activity cliff analysis
In this study, we developed predictive models for binary carcinogenicity data in rats using a range of feature-based and chemical language modeling approaches.
- Open Access
- 2 January 2026
- Published by American Chemical Society (ACS) in ChemRxiv
A round-robin exercise for the precise prediction of aqueous solubility of organic chemicals using chemometric, machine learning, and stacking ensemble of deep learning models
Aqueous solubility is an important property for assessing the druggability and ecotoxicological effects of molecules. Successful drug candidates should have optimal aqueous solubility to improve bioavailability to target tissues. To effectively screen molecules in a short period of time, reliable predictive models are highly useful. In the present study, we conducted a round-robin exercise using a large, curated dataset of over 6000 compounds to predict aqueous solubility quantitatively. The six participating groups used an array of Machine Learning and Deep Learning algorithms to develop models with strong robustness and external predictive performance. All the models underwent rigorous Leave-One-Out and 10-fold cross-validation. The diversity of training sets and descriptor types used by different groups paved the way for exploring the mechanistic basis for the efficient identification of contributing features. The best-performing model was selected using the statistical Sum of Ranking Differences (SRD) approach, considering the performances on training, cross-validation, and test, as well as the performance difference between the training and test sets. Additionally, a curated, true external set was screened by the six different models. Here, the best-performing model was selected using a consensus ranking strategy based on Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R_Ext^2. In both approaches, i.e., the inherent model performance in terms of training, test, and cross-validation statistics, and the ability of the model to efficiently predict true external data, the Stacking Ensemble of Deep q-RASPR model emerged as the winner. This model showed comparable predictive performance to the previously reported model, which apparently lacked a proper data curation workflow and contained a significant number of duplicates and mixtures in its dataset, which can inflate model statistics. The insights from the different feature contributions from the different groups identified the useful structural and physicochemical aspects, which can help synthetic chemists to optimize molecules.
- 2 January 2026
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessA round-robin exercise for the precise prediction of aqueous solubility of organic chemicals using chemometric, machine learning, and stacking ensemble of deep learning models
Aqueous solubility is an important property for assessing the druggability and ecotoxicological effects of molecules. Successful drug candidates should have optimal aqueous solubility to improve bioavailability to target tissues. To effectively screen molecules in a short period of time, reliable predictive models are highly useful. In the present study, we conducted a round-robin exercise using a large, curated dataset of over 6000 compounds to predict aqueous solubility quantitatively. The six participating groups used an array of Machine Learning and Deep Learning algorithms to develop models with strong robustness and external predictive performance. All the models underwent rigorous Leave-One-Out and 10-fold cross-validation. The diversity of training sets and descriptor types used by different groups paved the way for exploring the mechanistic basis for the efficient identification of contributing features. The best-performing model was selected using the statistical Sum of Ranking Differences (SRD) approach, considering the performances on training, cross-validation, and test, as well as the performance difference between the training and test sets. Additionally, a curated, true external set was screened by the six different models. Here, the best-performing model was selected using a consensus ranking strategy based on Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and R_Ext^2. In both approaches, i.e., the inherent model performance in terms of training, test, and cross-validation statistics, and the ability of the model to efficiently predict true external data, the Stacking Ensemble of Deep q-RASPR model emerged as the winner. This model showed comparable predictive performance to the previously reported model, which apparently lacked a proper data curation workflow and contained a significant number of duplicates and mixtures in its dataset, which can inflate model statistics. The insights from the different feature contributions from the different groups identified the useful structural and physicochemical aspects, which can help synthetic chemists to optimize molecules.
- 12 November 2025
- Published by Wiley in WIREs Computational Molecular Science
- Vol. 15 (6)
From Feature‐Based Chemical Similarity to Chemical Language Models—A Paradigm Shift in Computer‐Aided Molecular Design and Property Predictions
This review provides a comprehensive overview of the paradigm shift for computer‐aided molecular design and property predictions from similarity‐based modeling, including quantitative structure–activity/property relationship (QSAR/QSPR), read‐across, read‐across structure–activity relationship (RASAR), and pharmacophore mapping to sequence‐based chemical language models (CLMs) using deep learning techniques. Starting with multiple methods of chemical structure and latent chemical space representations and touching the molecular descriptor‐ and fingerprint‐based classical type modeling, this review introduces string‐based deep learning models involving techniques like recurrent neural networks (RNNs) with long short‐term memory (LSTM) and other architectures such as variational autoencoder (VAE), attention models, and generative adversarial networks (GANs). The basics of more efficient transformer models are also discussed. The problem‐solving of training with scarce data using transfer learning, data augmentation, and natural‐product‐inspired training is analyzed. The applications of CLMs in the de novo design of small molecules of medicinal interest, enzymes, peptides, and multitask agents, the predictions of properties of drug candidates, and activity cliffs are presented. The applications of CLMs in materials science and predictive toxicology are also mentioned. We discuss the limitations of feature‐based modeling approaches confined to a restricted feature space. In contrast, CLMs lack specific insights into aspects like SARs, bioisosteric replacements, synthesizability, and so forth, which collectively hinder their regulatory acceptance and acceptance by synthetic chemists. This review concludes that cheminformaticians need to utilize two complementary approaches, where factors like simplicity, reproducibility, and regulatory acceptability may prompt the use of feature‐based approaches while aiming for higher accuracy and generating novel molecules may drive toward adopting CLMs. This article is categorized under: Data Science > Chemoinformatics Structure and Mechanism > Computational Biochemistry and Biophysics Software > Molecular Modeling
- 12 November 2025
- Published by Wiley in WIREs Computational Molecular Science
- Vol. 15 (6)
From Feature‐Based Chemical Similarity to Chemical Language Models—A Paradigm Shift in Computer‐Aided Molecular Design and Property Predictions
This review provides a comprehensive overview of the paradigm shift for computer‐aided molecular design and property predictions from similarity‐based modeling, including quantitative structure–activity/property relationship (QSAR/QSPR), read‐across, read‐across structure–activity relationship (RASAR), and pharmacophore mapping to sequence‐based chemical language models (CLMs) using deep learning techniques. Starting with multiple methods of chemical structure and latent chemical space representations and touching the molecular descriptor‐ and fingerprint‐based classical type modeling, this review introduces string‐based deep learning models involving techniques like recurrent neural networks (RNNs) with long short‐term memory (LSTM) and other architectures such as variational autoencoder (VAE), attention models, and generative adversarial networks (GANs). The basics of more efficient transformer models are also discussed. The problem‐solving of training with scarce data using transfer learning, data augmentation, and natural‐product‐inspired training is analyzed. The applications of CLMs in the de novo design of small molecules of medicinal interest, enzymes, peptides, and multitask agents, the predictions of properties of drug candidates, and activity cliffs are presented. The applications of CLMs in materials science and predictive toxicology are also mentioned. We discuss the limitations of feature‐based modeling approaches confined to a restricted feature space. In contrast, CLMs lack specific insights into aspects like SARs, bioisosteric replacements, synthesizability, and so forth, which collectively hinder their regulatory acceptance and acceptance by synthetic chemists. This review concludes that cheminformaticians need to utilize two complementary approaches, where factors like simplicity, reproducibility, and regulatory acceptability may prompt the use of feature‐based approaches while aiming for higher accuracy and generating novel molecules may drive toward adopting CLMs. This article is categorized under: Data Science > Chemoinformatics Structure and Mechanism > Computational Biochemistry and Biophysics Software > Molecular Modeling
- Open Access
- 28 May 2025
- Published by American Chemical Society (ACS) in ChemRxiv
A New Approach Methodology (NAM) for carcinogenicity prediction of organic chemicals using the Multiclass ARKA framework and machine-learning-based stacking regression
The continuous accumulation of agricultural and industrial chemicals in the environment has significantly impacted the flora and fauna, resulting in disruptions in food chains and disturbances in the biological ecosystem. A high fraction of such chemicals poses a wide array of health risks for humans by affecting various adverse outcome pathways (AOPs). Carcinogenicity has been one of the most alarming adverse effects exhibited by these chemicals, which affects millions worldwide. For the efficient identification and safe disposal of such agricultural and industrial chemicals, it becomes necessary to quickly and easily determine toxicity and fill toxicity data gaps. In this study, we have developed multiple Machine Learning (ML) regression-based models that quantitatively predict the Oral Slope Factor (OSF) and Inhalation Slope Factor (ISF) of chemicals, identifying and prioritizing carcinogenicity risks. At first, we have developed Partial Least Squares (PLS) Quantitative Structure-Activity Relationship (QSAR) models. Aiming to further enhance the robustness and external predictivity of the models and better utilize the available chemical space, we have developed similarity-driven quantitative Read-Across Structure-Activity Relationship (q-RASAR) models. Further, we explored the Arithmetic Residuals in K-Groups Analysis (ARKA) to develop the Hybrid ARKA and ARKA-RASAR models in the quest for considering the response range-specific contribution of descriptors. We have used simple and reproducible Partial Least Squares (PLS) modeling algorithm to develop QSAR, q-RASAR, Hybrid ARKA and ARKA-RASAR models for both the responses, and further applied a wide array of ML modeling algorithms like Linear Support Vector Regression (LSVR), Ridge Regression (RR), k-Nearest Neighbor Regression (k-NN), Multilayer Perceptron Regression (MLP), Random Forests Regression (RF), Extra Trees Regression (ET), Gradient Boosting Regression (GB), PLS, and Multiple Linear Regression (MLR) as stacking regressors. The best performing models were selected using the multi-criteria decision-making approach – the Sum of Ranking Differences (SRD), considering training, test, and cross-validation statistics. Additionally, we have predicted the OSF and ISF of a true external data set and showed that the quantitative results align well with the reported carcinogenic status. With enhanced robustness and external predictivity of the models, the ARKA-RASAR approach has been shown to be a useful tool in ecotoxicological risk assessments.
- 28 May 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessA New Approach Methodology (NAM) for carcinogenicity prediction of organic chemicals using the Multiclass ARKA framework and machine-learning-based stacking regression
The continuous accumulation of agricultural and industrial chemicals in the environment has significantly impacted the flora and fauna, resulting in disruptions in food chains and disturbances in the biological ecosystem. A high fraction of such chemicals poses a wide array of health risks for humans by affecting various adverse outcome pathways (AOPs). Carcinogenicity has been one of the most alarming adverse effects exhibited by these chemicals, which affects millions worldwide. For the efficient identification and safe disposal of such agricultural and industrial chemicals, it becomes necessary to quickly and easily determine toxicity and fill toxicity data gaps. In this study, we have developed multiple Machine Learning (ML) regression-based models that quantitatively predict the Oral Slope Factor (OSF) and Inhalation Slope Factor (ISF) of chemicals, identifying and prioritizing carcinogenicity risks. At first, we have developed Partial Least Squares (PLS) Quantitative Structure-Activity Relationship (QSAR) models. Aiming to further enhance the robustness and external predictivity of the models and better utilize the available chemical space, we have developed similarity-driven quantitative Read-Across Structure-Activity Relationship (q-RASAR) models. Further, we explored the Arithmetic Residuals in K-Groups Analysis (ARKA) to develop the Hybrid ARKA and ARKA-RASAR models in the quest for considering the response range-specific contribution of descriptors. We have used simple and reproducible Partial Least Squares (PLS) modeling algorithm to develop QSAR, q-RASAR, Hybrid ARKA and ARKA-RASAR models for both the responses, and further applied a wide array of ML modeling algorithms like Linear Support Vector Regression (LSVR), Ridge Regression (RR), k-Nearest Neighbor Regression (k-NN), Multilayer Perceptron Regression (MLP), Random Forests Regression (RF), Extra Trees Regression (ET), Gradient Boosting Regression (GB), PLS, and Multiple Linear Regression (MLR) as stacking regressors. The best performing models were selected using the multi-criteria decision-making approach – the Sum of Ranking Differences (SRD), considering training, test, and cross-validation statistics. Additionally, we have predicted the OSF and ISF of a true external data set and showed that the quantitative results align well with the reported carcinogenic status. With enhanced robustness and external predictivity of the models, the ARKA-RASAR approach has been shown to be a useful tool in ecotoxicological risk assessments.
- 3 April 2025
- Published by Springer Nature
- p. 167-190
Applications of Quantitative Read-Across Structure–Property Relationship (q-RASPR) Modeling in the Field of Materials Science
- 3 April 2025
- Published by Springer Nature
- p. 167-190
Applications of Quantitative Read-Across Structure–Property Relationship (q-RASPR) Modeling in the Field of Materials Science