Sandip Patel
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sandip Patel has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Viral Patel
University Of Cumberlands, Williamsburg, 40769, KY, US
2 shared publications - Albert Price
International AIDs Vaccine Initiative (IAVI), 125 Broad Street, 9th Floor, New York, NY 10004, USA
1 shared publication - Daniel Craig
International AIDs Vaccine Initiative (IAVI), 125 Broad Street, 9th Floor, New York, NY 10004, USA
1 shared publication - Eu-Min Kim
SK bioscience Co., Ltd., 38, Yeongudanji-ro, Yeonsu-gu, Incheon 13493, Republic of Korea
1 shared publication - Jinhwan Shin
SK bioscience Co., Ltd., 38, Yeongudanji-ro, Yeonsu-gu, Incheon 13493, Republic of Korea
1 shared publication - Jun-Hong Jang
SK bioscience Co., Ltd., 38, Yeongudanji-ro, Yeonsu-gu, Incheon 13493, Republic of Korea
1 shared publication
Latest Publications
- Open Access
- 3 September 2026
- Published by MDPI AG in Pharmaceutics
- Vol. 18 (9) , 1110
Scalable Production of a De Novo SARS-CoV-2 Antiviral Miniprotein in Escherichia coli
Background/Objectives: The rapid emergence of SARS-CoV-2 variants that evade neutralizing antibodies underscores the need for new antiviral biologics that integrate precise molecular design with scalable, cost-effective manufacturing. Computationally designed miniproteins targeting the spike protein’s receptor-binding domain (RBD) offer a promising alternative to monoclonal antibodies because of their small size, high thermal stability, and compatibility with microbial expression systems. Here, we detail the full development and cGMP manufacturing of IPD-52520, a novel antiviral miniprotein produced using an optimized E. coli platform. Methods: Two candidate miniproteins, a homotrimeric construct (Trimer, IPD-52520, 17 kDa) and a tandem fusion construct (Daisy, IPD-52521, 25 kDa), were evaluated in parallel through systematic optimization of strain selection, media formulation, fed-batch fermentation, inclusion-body solubilization, refolding, and chromatographic purification. The leading candidate was scaled from 5 L to 50 L under cGMP conditions, and biophysical and stability analyses were conducted to support nonclinical and Phase 1 clinical development. Results: The Trimer was selected as the lead molecule based on superior preclinical efficacy, favorable pharmacokinetics, and higher volumetric yields. The optimized process yields about 2 g/L of purified protein at over 90% purity. Scale-up from 5 L to 50 L demonstrated excellent batch consistency across six independent runs. Biophysical studies confirmed a well-folded, predominantly alpha-helical trimer (Tm = 73.4 °C; polydispersity = 1.005) with an intact primary structure and strong target binding (KD < 1 pM). Stability studies show the drug remains stable at 2–8 °C for at least 12 months, with ongoing long-term studies. Conclusions: These findings demonstrate that computationally designed antiviral miniproteins can be translated into scalable biologics and establish a platform for rapid therapeutic development against current and future pandemics.
- 3 September 2026
- Published by MDPI AG in Pharmaceutics
- Vol. 18 (9) , 1110
Open AccessScalable Production of a De Novo SARS-CoV-2 Antiviral Miniprotein in Escherichia coli
Background/Objectives: The rapid emergence of SARS-CoV-2 variants that evade neutralizing antibodies underscores the need for new antiviral biologics that integrate precise molecular design with scalable, cost-effective manufacturing. Computationally designed miniproteins targeting the spike protein’s receptor-binding domain (RBD) offer a promising alternative to monoclonal antibodies because of their small size, high thermal stability, and compatibility with microbial expression systems. Here, we detail the full development and cGMP manufacturing of IPD-52520, a novel antiviral miniprotein produced using an optimized E. coli platform. Methods: Two candidate miniproteins, a homotrimeric construct (Trimer, IPD-52520, 17 kDa) and a tandem fusion construct (Daisy, IPD-52521, 25 kDa), were evaluated in parallel through systematic optimization of strain selection, media formulation, fed-batch fermentation, inclusion-body solubilization, refolding, and chromatographic purification. The leading candidate was scaled from 5 L to 50 L under cGMP conditions, and biophysical and stability analyses were conducted to support nonclinical and Phase 1 clinical development. Results: The Trimer was selected as the lead molecule based on superior preclinical efficacy, favorable pharmacokinetics, and higher volumetric yields. The optimized process yields about 2 g/L of purified protein at over 90% purity. Scale-up from 5 L to 50 L demonstrated excellent batch consistency across six independent runs. Biophysical studies confirmed a well-folded, predominantly alpha-helical trimer (Tm = 73.4 °C; polydispersity = 1.005) with an intact primary structure and strong target binding (KD < 1 pM). Stability studies show the drug remains stable at 2–8 °C for at least 12 months, with ongoing long-term studies. Conclusions: These findings demonstrate that computationally designed antiviral miniproteins can be translated into scalable biologics and establish a platform for rapid therapeutic development against current and future pandemics.
- Open Access
Machine Learning-Enabled Raman Spectroscopy for Process Analytical Technology and Real-Time Release Testing in Bioprocess Manufacturing: A Comparative Predictive Modeling Study
Analytical technologies that can provide quick, precise, and continuous information regarding process performance are necessary for the development of biopharmaceutical manufacturing. Conventional bioprocess monitoring is largely dependent on laboratory-based data and offline sampling, which can restrict process management and cause delays in decision-making. This study develops a machine learning-enabled Raman spectroscopy framework for Process Analytical Technology (PAT) and Real-Time Release Testing (RTRT) applications in bioprocess manufacturing. Five predictive modeling techniques Partial Least Squares (PLS) regression, Support Vector Regression (SVR), Random Forest, Extreme Gradient Boosting (XGBoost), and Neural Networks were used to analyze Raman spectral data from an Escherichia coli fermentation dataset. The models were assessed using the coefficient of determination (R2), root mean square error (RMSE), and mean absolute error (MAE) to predict two crucial fermentation parameters: the concentrations of glucose and acetate. The superior performance of PLS regression for glucose prediction and the improved prediction accuracy of XGBoost for acetate concentration demonstrated the importance of selecting modeling techniques based on biological complexity. Explainable artificial intelligence using SHAP analysis was incorporated to improve model transparency by identifying Raman spectral regions contributing to predictions. The suggested architecture shows how Raman spectroscopy and machine learning can be combined to assist automated process monitoring, enhance process comprehension, and hasten the implementation of real-time quality judgments in next-generation biomanufacturing.
Open AccessMachine Learning-Enabled Raman Spectroscopy for Process Analytical Technology and Real-Time Release Testing in Bioprocess Manufacturing: A Comparative Predictive Modeling Study
Analytical technologies that can provide quick, precise, and continuous information regarding process performance are necessary for the development of biopharmaceutical manufacturing. Conventional bioprocess monitoring is largely dependent on laboratory-based data and offline sampling, which can restrict process management and cause delays in decision-making. This study develops a machine learning-enabled Raman spectroscopy framework for Process Analytical Technology (PAT) and Real-Time Release Testing (RTRT) applications in bioprocess manufacturing. Five predictive modeling techniques Partial Least Squares (PLS) regression, Support Vector Regression (SVR), Random Forest, Extreme Gradient Boosting (XGBoost), and Neural Networks were used to analyze Raman spectral data from an Escherichia coli fermentation dataset. The models were assessed using the coefficient of determination (R2), root mean square error (RMSE), and mean absolute error (MAE) to predict two crucial fermentation parameters: the concentrations of glucose and acetate. The superior performance of PLS regression for glucose prediction and the improved prediction accuracy of XGBoost for acetate concentration demonstrated the importance of selecting modeling techniques based on biological complexity. Explainable artificial intelligence using SHAP analysis was incorporated to improve model transparency by identifying Raman spectral regions contributing to predictions. The suggested architecture shows how Raman spectroscopy and machine learning can be combined to assist automated process monitoring, enhance process comprehension, and hasten the implementation of real-time quality judgments in next-generation biomanufacturing.
- Open Access
- 1 July 2026
- Published by Elsevier in SSRN Electronic Journal
Time-Temperature-Aware Multi-Output Machine Learning Framework for Predicting Viscosity, Aggregation, Stability, and Bioactivity Decay in High-Concentration Monoclonal Antibody Formulations
Elevated protein concentrations can lead to increased viscosity, aggregation, and degradation during storage, making the creation of high-concentration monoclonal antibody (mAb) formulations difficult. Traditional formulation development takes a lot of time and resources because it depends on long-term stability studies and intensive experimental screening. A time-temperature-aware multi-output machine learning system that can concurrently predict viscosity, monomer retention, aggregation growth, and monomer loss under various storage circumstances is shown in this work. Formulation parameters such as additive identity, additive concentration, monoclonal antibody concentration, diffusion interaction parameter (kD), second virial coefficient (B22), and polydispersity index (PDI) were mixed with storage temperature and time. A longitudinal, storage-aware dataset of 232 observations-196 of which were full and kept for model development-was created from experimental stability data. To reduce formulation leakage and guarantee reliable evaluation, a Random Forest Multi-Output Regression architecture was used in conjunction with Group K-Fold validation. Strong predictive capacity for aggregation prediction (R² = 0.779) and moderate predictive performance for monomer retention (R² = 0.556) and monomer loss (R² = 0.552) was shown by cross-validation findings. An independent bioactivity-decay model with an R² of 0.885 was trained on 1,920 data. To score formulations according to anticipated long-term stability under accelerated storage circumstances, a composite stability scoring methodology was also created. The suggested approach shows how physicochemical characterization, storage-stress variables, machine-learning prediction, and formulation optimization can be integrated to promote Quality-by-Design initiatives and expedited biopharmaceutical formulation development.
- 1 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessTime-Temperature-Aware Multi-Output Machine Learning Framework for Predicting Viscosity, Aggregation, Stability, and Bioactivity Decay in High-Concentration Monoclonal Antibody Formulations
Elevated protein concentrations can lead to increased viscosity, aggregation, and degradation during storage, making the creation of high-concentration monoclonal antibody (mAb) formulations difficult. Traditional formulation development takes a lot of time and resources because it depends on long-term stability studies and intensive experimental screening. A time-temperature-aware multi-output machine learning system that can concurrently predict viscosity, monomer retention, aggregation growth, and monomer loss under various storage circumstances is shown in this work. Formulation parameters such as additive identity, additive concentration, monoclonal antibody concentration, diffusion interaction parameter (kD), second virial coefficient (B22), and polydispersity index (PDI) were mixed with storage temperature and time. A longitudinal, storage-aware dataset of 232 observations-196 of which were full and kept for model development-was created from experimental stability data. To reduce formulation leakage and guarantee reliable evaluation, a Random Forest Multi-Output Regression architecture was used in conjunction with Group K-Fold validation. Strong predictive capacity for aggregation prediction (R² = 0.779) and moderate predictive performance for monomer retention (R² = 0.556) and monomer loss (R² = 0.552) was shown by cross-validation findings. An independent bioactivity-decay model with an R² of 0.885 was trained on 1,920 data. To score formulations according to anticipated long-term stability under accelerated storage circumstances, a composite stability scoring methodology was also created. The suggested approach shows how physicochemical characterization, storage-stress variables, machine-learning prediction, and formulation optimization can be integrated to promote Quality-by-Design initiatives and expedited biopharmaceutical formulation development.