Saratram Gopalakrishnan
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
Research Subjects
Saratram Gopalakrishnan has a total of 106 co-authors. Here are the top co-authors by number of shared publications:
- Costas D. Maranas
Department of Chemical Engineering, The Pennsylvania State University, University Park, Pennsylvania, United States of America
11 shared publications - Nathan E. Lewis
University of Georgia
5 shared publications Cleo Kontoravdi
Imperial College London
3 shared publicationsElcin Icten
Amgen Inc
3 shared publications- Hadi Nazem-Bokaee
The John Curtin School of Medical Research, College of Science and Medicine, The Australian National University, Canberra ACT 2600, Australia
3 shared publications - James Gregory Ferry
Department of Biochemistry and Molecular Biology, Pennsylvania State University, University Park, PA 16801, USA
3 shared publications
Latest Publications
- Open Access
- 14 August 2025
- Published by Taylor & Francis in mAbs
- Vol. 17 (1) , 2547084
Machine learning-driven optimization of culture conditions and media components to mitigate charge heterogeneity in monoclonal antibody production: current advances and future perspectives
Charge heterogeneity in monoclonal antibodies (mAbs), caused by post-translational modifications, remains a substantial obstacle to ensuring consistent, stable, and effective therapeutics. Conventional optimization techniques, such as one-factor-at-a-time and design of experiments, often fail to capture the complex, nonlinear interactions between culture parameters (e.g. pH, temperature, duration) and medium components (e.g. glucose, metal ions, amino acids). This review highlights machine learning (ML) as a powerful approach for modeling these relationships and forecasting charge variant profiles in CHO cell-based mAb process development. We summarize supervised learning and regression methods used to link process conditions with charge heterogeneity and present case studies showing ML's role in reducing acidic and basic variants. We also discuss challenges related to data quality, model interpretability, scalability, and regulatory compliance. Finally, we propose a roadmap for adaptive, ML-driven optimization strategies for bioprocess development, aligned with Quality-by-Design principles.
- 14 August 2025
- Published by Taylor & Francis in mAbs
- Vol. 17 (1) , 2547084
Open AccessMachine learning-driven optimization of culture conditions and media components to mitigate charge heterogeneity in monoclonal antibody production: current advances and future perspectives
Charge heterogeneity in monoclonal antibodies (mAbs), caused by post-translational modifications, remains a substantial obstacle to ensuring consistent, stable, and effective therapeutics. Conventional optimization techniques, such as one-factor-at-a-time and design of experiments, often fail to capture the complex, nonlinear interactions between culture parameters (e.g. pH, temperature, duration) and medium components (e.g. glucose, metal ions, amino acids). This review highlights machine learning (ML) as a powerful approach for modeling these relationships and forecasting charge variant profiles in CHO cell-based mAb process development. We summarize supervised learning and regression methods used to link process conditions with charge heterogeneity and present case studies showing ML's role in reducing acidic and basic variants. We also discuss challenges related to data quality, model interpretability, scalability, and regulatory compliance. Finally, we propose a roadmap for adaptive, ML-driven optimization strategies for bioprocess development, aligned with Quality-by-Design principles.
- Open Access
A community-consensus reconstruction of Chinese Hamster metabolism enables structural systems biology analyses to decipher metabolic rewiring in lactate-free CHO cells
Genome-scale metabolic models (GEMs) are indispensable for studying and engineering cellular metabolism. Here, we present i CHO3K, a community-consensus, manually-curated reconstruction of the Chinese Hamster metabolic network. In addition to accounting for 11004 reactions associated with 3597 genes, i CHO3K includes 3489 protein structures and structural descriptors for >70% of its 7377 metabolites, enabling deeper exploration of the link between molecular structure and cellular metabolism. We used i CHO3K to contextualize transcriptomics and metabolomics data from a CHO cell line in which lactate secretion is abolished. We found the reduced glycolytic flux and enhanced TCA cycle flux were accompanied by an elevated NADH and PEP levels in these cells, consistent with experimental measurements. Leveraging i CHO3K’s structural annotations, we identified candidate binding interactions of NADH and PEP with glycolytic enzymes showing model-predicted differential flux, suggesting novel allosteric regulation associated with the observed decrease in glucose uptake and glycolysis. Overall, i CHO3K offers a valuable framework for systematic integration of omics data, improved flux predictions, and structure-guided insights, thus advancing CHO cell engineering and enhancing biomanufacturing efficiency.
Open AccessA community-consensus reconstruction of Chinese Hamster metabolism enables structural systems biology analyses to decipher metabolic rewiring in lactate-free CHO cells
Genome-scale metabolic models (GEMs) are indispensable for studying and engineering cellular metabolism. Here, we present i CHO3K, a community-consensus, manually-curated reconstruction of the Chinese Hamster metabolic network. In addition to accounting for 11004 reactions associated with 3597 genes, i CHO3K includes 3489 protein structures and structural descriptors for >70% of its 7377 metabolites, enabling deeper exploration of the link between molecular structure and cellular metabolism. We used i CHO3K to contextualize transcriptomics and metabolomics data from a CHO cell line in which lactate secretion is abolished. We found the reduced glycolytic flux and enhanced TCA cycle flux were accompanied by an elevated NADH and PEP levels in these cells, consistent with experimental measurements. Leveraging i CHO3K’s structural annotations, we identified candidate binding interactions of NADH and PEP with glycolytic enzymes showing model-predicted differential flux, suggesting novel allosteric regulation associated with the observed decrease in glucose uptake and glycolysis. Overall, i CHO3K offers a valuable framework for systematic integration of omics data, improved flux predictions, and structure-guided insights, thus advancing CHO cell engineering and enhancing biomanufacturing efficiency.
- Open Access
Multi-omic characterization of antibody-producing CHO cell lines elucidates metabolic reprogramming and nutrient uptake bottlenecks
Characterizing the phenotypic diversity and metabolic capabilities of industrially relevant manufacturing cell lines is critical to bioprocess optimization and cell line development. Metabolic capabilities of the production hosts limit nutrient and resource channeling into desired cellular processes and can have a profound impact on productivity but cannot be directly inferred from measured data such as spent media concentrations or transcriptomics. Here, we present an integrated multi-omic characterization approach combining exo-metabolomics, transcriptomics, and genome-scale metabolic network analysis and apply it to three antibody-producing Chinese Hamster Ovary cell lines to reprogramming features associated with high-producer clones and metabolic bottlenecks limiting product production in an industrial bioprocess. Analysis of individual datatypes revealed a decreased nitrogenous byproduct secretion in high-producing clones and the topological changes in peripheral metabolic pathway expression associated with phase shifts. An integrated omics analysis in the context of the genome-scale metabolic model elucidated the differences in central metabolism and identified amino acid utilization bottlenecks limiting cell growth and antibody production that were not evident from exo-metabolomics or transcriptomics alone. Thus, we demonstrate the utility of a multi-omics characterization in providing an in-depth understanding of cellular metabolism, which is critical to efforts in cell engineering and bioprocess optimization.
Open AccessMulti-omic characterization of antibody-producing CHO cell lines elucidates metabolic reprogramming and nutrient uptake bottlenecks
Characterizing the phenotypic diversity and metabolic capabilities of industrially relevant manufacturing cell lines is critical to bioprocess optimization and cell line development. Metabolic capabilities of the production hosts limit nutrient and resource channeling into desired cellular processes and can have a profound impact on productivity but cannot be directly inferred from measured data such as spent media concentrations or transcriptomics. Here, we present an integrated multi-omic characterization approach combining exo-metabolomics, transcriptomics, and genome-scale metabolic network analysis and apply it to three antibody-producing Chinese Hamster Ovary cell lines to reprogramming features associated with high-producer clones and metabolic bottlenecks limiting product production in an industrial bioprocess. Analysis of individual datatypes revealed a decreased nitrogenous byproduct secretion in high-producing clones and the topological changes in peripheral metabolic pathway expression associated with phase shifts. An integrated omics analysis in the context of the genome-scale metabolic model elucidated the differences in central metabolism and identified amino acid utilization bottlenecks limiting cell growth and antibody production that were not evident from exo-metabolomics or transcriptomics alone. Thus, we demonstrate the utility of a multi-omics characterization in providing an in-depth understanding of cellular metabolism, which is critical to efforts in cell engineering and bioprocess optimization.
- Open Access
COSMIC-dFBA: A novel multi-scale hybrid framework for bioprocess modeling
Metabolism governs cell performance in biomanufacturing, as it fuels growth and productivity. However, even in well-controlled culture systems, metabolism is dynamic, with shifting objectives and resources, thus limiting the predictive capability of mechanistic models for process design and optimization. Here, we present Cellular Objectives and State Modulation In bioreaCtors (COSMIC)-dFBA, a hybrid multi-scale modeling paradigm that accurately predicts cell density, antibody titer, and bioreactor metabolite concentration profiles. Using machine-learning, COSMIC-dFBA decomposes the instantaneous metabolite uptake and secretion rates in a bioreactor into weighted contributions from each cell state (growth or antibody-producing state) and integrates these with a genome-scale metabolic model. A major strength of COSMIC-dFBA is that it can be parameterized with only metabolite concentrations from spent media, although constraining the metabolic model with other omics data can further improve its capabilities. Using COSMIC-dFBA, we can predict the final cell density and antibody titer to within 10% of the measured data, and compared to a standard dFBA model, we found the framework showed a 90% and 72% improvement in cell density and antibody titer prediction, respectively. Thus, we demonstrate our hybrid modeling framework effectively captures cellular metabolism and expands the applicability of dFBA to model the dynamic conditions in a bioreactor.
Open AccessCOSMIC-dFBA: A novel multi-scale hybrid framework for bioprocess modeling
Metabolism governs cell performance in biomanufacturing, as it fuels growth and productivity. However, even in well-controlled culture systems, metabolism is dynamic, with shifting objectives and resources, thus limiting the predictive capability of mechanistic models for process design and optimization. Here, we present Cellular Objectives and State Modulation In bioreaCtors (COSMIC)-dFBA, a hybrid multi-scale modeling paradigm that accurately predicts cell density, antibody titer, and bioreactor metabolite concentration profiles. Using machine-learning, COSMIC-dFBA decomposes the instantaneous metabolite uptake and secretion rates in a bioreactor into weighted contributions from each cell state (growth or antibody-producing state) and integrates these with a genome-scale metabolic model. A major strength of COSMIC-dFBA is that it can be parameterized with only metabolite concentrations from spent media, although constraining the metabolic model with other omics data can further improve its capabilities. Using COSMIC-dFBA, we can predict the final cell density and antibody titer to within 10% of the measured data, and compared to a standard dFBA model, we found the framework showed a 90% and 72% improvement in cell density and antibody titer prediction, respectively. Thus, we demonstrate our hybrid modeling framework effectively captures cellular metabolism and expands the applicability of dFBA to model the dynamic conditions in a bioreactor.
- Open Access
Guidelines for extracting biologically relevant context-specific metabolic models using gene expression data
Genome-scale metabolic models comprehensively describe an organism’s metabolism and can be tailored using omics data to model condition-specific physiology. The quality of context-specific models is impacted by (i) choice of algorithm and parameters and (ii) alternate context-specific models that equally explain the -omics data. Here we quantify the influence of alternate optima on microbial and mammalian model extraction using GIMME, iMAT, MBA, and mCADRE. We find that metabolic tasks defining an organism’s phenotype must be explicitly and quantitatively protected. The scope of alternate models is strongly influenced by algorithm choice and the topological properties of the parent genome-scale model with fatty acid metabolism and intracellular metabolite transport contributing much to alternate solutions in all models. mCADRE extracted the most reproducible context-specific models and models generated using MBA had the most alternate solutions. There were fewer qualitatively different solutions generated by GIMME in E. coli , but these increased substantially in the mammalian models. Screening ensembles using a receiver operating characteristic plot identified the best-performing models. A comprehensive evaluation of models extracted using combinations of extraction methods and expression thresholds revealed that GIMME generated the best-performing models in E. coli , whereas mCADRE is better suited for complex mammalian models. These findings suggest guidelines for benchmarking -omics integration algorithms and motivate the development of a systematic workflow to enumerate alternate models and extract biologically relevant context-specific models.
Open AccessGuidelines for extracting biologically relevant context-specific metabolic models using gene expression data
Genome-scale metabolic models comprehensively describe an organism’s metabolism and can be tailored using omics data to model condition-specific physiology. The quality of context-specific models is impacted by (i) choice of algorithm and parameters and (ii) alternate context-specific models that equally explain the -omics data. Here we quantify the influence of alternate optima on microbial and mammalian model extraction using GIMME, iMAT, MBA, and mCADRE. We find that metabolic tasks defining an organism’s phenotype must be explicitly and quantitatively protected. The scope of alternate models is strongly influenced by algorithm choice and the topological properties of the parent genome-scale model with fatty acid metabolism and intracellular metabolite transport contributing much to alternate solutions in all models. mCADRE extracted the most reproducible context-specific models and models generated using MBA had the most alternate solutions. There were fewer qualitatively different solutions generated by GIMME in E. coli , but these increased substantially in the mammalian models. Screening ensembles using a receiver operating characteristic plot identified the best-performing models. A comprehensive evaluation of models extracted using combinations of extraction methods and expression thresholds revealed that GIMME generated the best-performing models in E. coli , whereas mCADRE is better suited for complex mammalian models. These findings suggest guidelines for benchmarking -omics integration algorithms and motivate the development of a systematic workflow to enumerate alternate models and extract biologically relevant context-specific models.
- Open Access
- 1 January 2022
- Published by Elsevier in Metabolic Engineering
- Vol. 69, 286-301
Assessing the impact of substrate-level enzyme regulations limiting ethanol titer in Clostridium thermocellum using a core kinetic model
- 1 January 2022
- Published by Elsevier in Metabolic Engineering
- Vol. 69, 286-301
Open AccessAssessing the impact of substrate-level enzyme regulations limiting ethanol titer in Clostridium thermocellum using a core kinetic model
- 1 December 2020
- Published by Elsevier in Current Opinion in Chemical Engineering
- Vol. 30, 17-25
Metabolic flux analysis reaching genome wide coverage: lessons learned and future perspectives
- 1 December 2020
- Published by Elsevier in Current Opinion in Chemical Engineering
- Vol. 30, 17-25
Metabolic flux analysis reaching genome wide coverage: lessons learned and future perspectives
- Open Access
- 15 October 2020
- Published by American Chemical Society (ACS)
13C Metabolic flux analysis and kinetic modeling of Clostridium thermocellum grown on cellobiose
- 15 October 2020
- Published by American Chemical Society (ACS)
Open Access13C Metabolic flux analysis and kinetic modeling of Clostridium thermocellum grown on cellobiose
- Open Access
- 10 September 2019
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 15 (9) , e1007319
From Escherichia coli mutant 13C labeling data to a core kinetic model: A kinetic model parameterization pipeline
Kinetic models of metabolic networks offer the promise of quantitative phenotype prediction. The mechanistic characterization of enzyme catalyzed reactions allows for tracing the effect of perturbations in metabolite concentrations and reaction fluxes in response to genetic and environmental perturbation that are beyond the scope of stoichiometric models. In this study, we develop a two-step computational pipeline for the rapid parameterization of kinetic models of metabolic networks using a curated metabolic model and available 13C-labeling distributions under multiple genetic and environmental perturbations. The first step involves the elucidation of all intracellular fluxes in a core model of E. coli containing 74 reactions and 61 metabolites using 13C-Metabolic Flux Analysis (13C-MFA). Here, fluxes corresponding to the mid-exponential growth phase are elucidated for seven single gene deletion mutants from upper glycolysis, pentose phosphate pathway and the Entner-Doudoroff pathway. The computed flux ranges are then used to parameterize the same (i.e., k-ecoli74) core kinetic model for E. coli with 55 substrate-level regulations using the newly developed K-FIT parameterization algorithm. The K-FIT algorithm employs a combination of equation decomposition and iterative solution techniques to evaluate steady-state fluxes in response to genetic perturbations. k-ecoli74 predicted 86% of flux values for strains used during fitting within a single standard deviation of 13C-MFA estimated values. By performing both tasks using the same network, errors associated with lack of congruity between the two networks are avoided, allowing for seamless integration of data with model building. Product yield predictions and comparison with previously developed kinetic models indicate shifts in flux ranges and the presence or absence of mutant strains delivering flux towards pathways of interest from training data significantly impact predictive capabilities. Using this workflow, the impact of completeness of fluxomic datasets and the importance of specific genetic perturbations on uncertainties in kinetic parameter estimation are evaluated. Microbial production hosts are used for production of a wide range of commodity chemicals. Improving the conversion efficiency of microbial strains is critical to the economic viability and the continued push towards the use of environmentally neutral bioprocesses as a means for producing the chemicals society depends on. Metabolic models have played a key role in helping us to predict metabolic behavior in response to environmental and genetic perturbation that can maximize efficiency. Recently, kinetic models of metabolism have re-emerged as a means for characterizing metabolism, offering improvements over their stoichiometric counterparts in both the type of information that can be gleaned from them, and in prediction accuracy. Despite recent developments, a lack of raw experimental data needed for flux elucidation and, subsequently, kinetic parameterization, and high computation cost have prevented the development of a uniform workflow for construction of the most informative kinetic models. Here, we have incorporated raw 13C-isotopic labeling data and a computationally inexpensive parameterization algorithm into a kinetic parameterization pipeline to ensure that the resulting kinetic model (k-ecoli74) conforms to experimental data. We show how the use of an identical metabolic network for flux elucidation and kinetic parameterization influences predictive capabilities.
- 10 September 2019
- Published by Public Library of Science (PLoS) in PLOS Computational Biology
- Vol. 15 (9) , e1007319
Open AccessFrom Escherichia coli mutant 13C labeling data to a core kinetic model: A kinetic model parameterization pipeline
Kinetic models of metabolic networks offer the promise of quantitative phenotype prediction. The mechanistic characterization of enzyme catalyzed reactions allows for tracing the effect of perturbations in metabolite concentrations and reaction fluxes in response to genetic and environmental perturbation that are beyond the scope of stoichiometric models. In this study, we develop a two-step computational pipeline for the rapid parameterization of kinetic models of metabolic networks using a curated metabolic model and available 13C-labeling distributions under multiple genetic and environmental perturbations. The first step involves the elucidation of all intracellular fluxes in a core model of E. coli containing 74 reactions and 61 metabolites using 13C-Metabolic Flux Analysis (13C-MFA). Here, fluxes corresponding to the mid-exponential growth phase are elucidated for seven single gene deletion mutants from upper glycolysis, pentose phosphate pathway and the Entner-Doudoroff pathway. The computed flux ranges are then used to parameterize the same (i.e., k-ecoli74) core kinetic model for E. coli with 55 substrate-level regulations using the newly developed K-FIT parameterization algorithm. The K-FIT algorithm employs a combination of equation decomposition and iterative solution techniques to evaluate steady-state fluxes in response to genetic perturbations. k-ecoli74 predicted 86% of flux values for strains used during fitting within a single standard deviation of 13C-MFA estimated values. By performing both tasks using the same network, errors associated with lack of congruity between the two networks are avoided, allowing for seamless integration of data with model building. Product yield predictions and comparison with previously developed kinetic models indicate shifts in flux ranges and the presence or absence of mutant strains delivering flux towards pathways of interest from training data significantly impact predictive capabilities. Using this workflow, the impact of completeness of fluxomic datasets and the importance of specific genetic perturbations on uncertainties in kinetic parameter estimation are evaluated. Microbial production hosts are used for production of a wide range of commodity chemicals. Improving the conversion efficiency of microbial strains is critical to the economic viability and the continued push towards the use of environmentally neutral bioprocesses as a means for producing the chemicals society depends on. Metabolic models have played a key role in helping us to predict metabolic behavior in response to environmental and genetic perturbation that can maximize efficiency. Recently, kinetic models of metabolism have re-emerged as a means for characterizing metabolism, offering improvements over their stoichiometric counterparts in both the type of information that can be gleaned from them, and in prediction accuracy. Despite recent developments, a lack of raw experimental data needed for flux elucidation and, subsequently, kinetic parameterization, and high computation cost have prevented the development of a uniform workflow for construction of the most informative kinetic models. Here, we have incorporated raw 13C-isotopic labeling data and a computationally inexpensive parameterization algorithm into a kinetic parameterization pipeline to ensure that the resulting kinetic model (k-ecoli74) conforms to experimental data. We show how the use of an identical metabolic network for flux elucidation and kinetic parameterization influences predictive capabilities.
- Open Access
- 14 December 2018
- Published by Oxford University Press (OUP) in Plant Physiology
- Vol. 179 (2) , 761-769
Genome-Scale Fluxome of Synechococcus elongatus UTEX 2973 Using Transient 13C-Labeling Data
Synechococcus elongatus UTEX 2973 (Synechococcus 2973) has the shortest reported doubling time (2.1 h) among cyanobacteria, making it a promising platform for the solar-based production of biochemicals. In this meta-analysis, its intracellular flux distribution was recomputed using genome-scale isotopic nonstationary 13C-metabolic flux analysis given the labeling dynamics of 13 metabolites reported in an earlier study. To achieve this, a genome-scale mapping model, namely imSyu593, was constructed using the imSyn617 mapping model for Synechocystis sp. PCC 6803 (Synechocystis 6803) as the starting point encompassing 593 reactions. The flux elucidation revealed nearly complete conversion (greater than 96%) of the assimilated carbon into biomass in Synechococcus 2973. In contrast, Synechocystis 6803 achieves complete conversion of only 86% of the assimilated carbon. This high biomass yield was enabled by the reincorporation of the fixed carbons lost in anabolic and photorespiratory pathways in conjunction with flux rerouting through a nondecarboxylating reaction such as phosphoketolase. This reincorporation of lost CO2 sustains a higher flux through the photorespiratory C2 cycle that fully meets the glycine and serine demands for growth. In accordance with the high carbon efficiency drive, acetyl-coenzyme A was entirely produced using the carbon-efficient phosphoketolase pathway. Comparison of the Synechococcus 2973 flux map with that of Synechocystis 6803 revealed differences in the use of Calvin cycle and photorespiratory pathway reactions. The two species used different reactions for the synthesis of metabolites such as fructose-6-phosphate, glycine, sedoheptulose-7-phosphate, and Ser. These findings allude to a highly carbon-efficient metabolism alongside the fast carbon uptake rate in Synechococcus 2973, which explains its faster growth rate.
- 14 December 2018
- Published by Oxford University Press (OUP) in Plant Physiology
- Vol. 179 (2) , 761-769
Open AccessGenome-Scale Fluxome of Synechococcus elongatus UTEX 2973 Using Transient 13C-Labeling Data
Synechococcus elongatus UTEX 2973 (Synechococcus 2973) has the shortest reported doubling time (2.1 h) among cyanobacteria, making it a promising platform for the solar-based production of biochemicals. In this meta-analysis, its intracellular flux distribution was recomputed using genome-scale isotopic nonstationary 13C-metabolic flux analysis given the labeling dynamics of 13 metabolites reported in an earlier study. To achieve this, a genome-scale mapping model, namely imSyu593, was constructed using the imSyn617 mapping model for Synechocystis sp. PCC 6803 (Synechocystis 6803) as the starting point encompassing 593 reactions. The flux elucidation revealed nearly complete conversion (greater than 96%) of the assimilated carbon into biomass in Synechococcus 2973. In contrast, Synechocystis 6803 achieves complete conversion of only 86% of the assimilated carbon. This high biomass yield was enabled by the reincorporation of the fixed carbons lost in anabolic and photorespiratory pathways in conjunction with flux rerouting through a nondecarboxylating reaction such as phosphoketolase. This reincorporation of lost CO2 sustains a higher flux through the photorespiratory C2 cycle that fully meets the glycine and serine demands for growth. In accordance with the high carbon efficiency drive, acetyl-coenzyme A was entirely produced using the carbon-efficient phosphoketolase pathway. Comparison of the Synechococcus 2973 flux map with that of Synechocystis 6803 revealed differences in the use of Calvin cycle and photorespiratory pathway reactions. The two species used different reactions for the synthesis of metabolites such as fructose-6-phosphate, glycine, sedoheptulose-7-phosphate, and Ser. These findings allude to a highly carbon-efficient metabolism alongside the fast carbon uptake rate in Synechococcus 2973, which explains its faster growth rate.