Jinyeop Song
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
Jinyeop Song has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Hyewon Jeong
Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, USA
4 shared publications - Sarah Soyeon Oh
Institute for Global Engagement & Empowerment, Yonsei University, Seoul, Republic of Korea
4 shared publications - Minkyu Jeon
Department of Computer Science, Princeton University
3 shared publications - Bada Kang
Mo–Im Kim Nursing Research Institute, Yonsei University College of Nursing, Seoul, Republic of Korea
2 shared publications - Bo-Seong Jeong
Department of Biological Sciences, KAIST Institute for the Biocentury, Korea Advanced Institute of Science and Technology
2 shared publications - Boyu Ren
Laboratory for Psychiatric Biostatistics, McLean Hospital , Belmont, MA 02478
2 shared publications
Latest Publications
- Open Access
Agentic campaign control for high-throughput de novo binder design
Progress in artificial intelligence has produced a rapidly growing ecosystem of methods for de novo protein design. With access to many specialized and often complementary tools, how does one use them effectively, especially with a finite compute budget? Here, we introduce Target-adaptive Rescue–Explore–eXploit (T-REX) , an agentic protein binder design framework that orchestrates multiple state-of-the-art protein generative models and structure evaluators. Over the course of a design campaign, T-REX leverages large language model (LLM) agents to reason over accumulating outcomes and decide whether to rescue promising candidates, explore alternative settings, or exploit productive routes, while a deterministic controller validates and schedules proposed actions. Across seven targets, we find that T-REX adaptively allocates its compute across methods in a target-dependent manner and achieves the highest throughput of structurally distinct hits compared to all baselines. These results suggest that adaptive orchestration can complement increasingly powerful protein design models, and we openly release T-REX to support compute-efficient binder design workflows.
Open AccessAgentic campaign control for high-throughput de novo binder design
Progress in artificial intelligence has produced a rapidly growing ecosystem of methods for de novo protein design. With access to many specialized and often complementary tools, how does one use them effectively, especially with a finite compute budget? Here, we introduce Target-adaptive Rescue–Explore–eXploit (T-REX) , an agentic protein binder design framework that orchestrates multiple state-of-the-art protein generative models and structure evaluators. Over the course of a design campaign, T-REX leverages large language model (LLM) agents to reason over accumulating outcomes and decide whether to rescue promising candidates, explore alternative settings, or exploit productive routes, while a deterministic controller validates and schedules proposed actions. Across seven targets, we find that T-REX adaptively allocates its compute across methods in a target-dependent manner and achieves the highest throughput of structurally distinct hits compared to all baselines. These results suggest that adaptive orchestration can complement increasingly powerful protein design models, and we openly release T-REX to support compute-efficient binder design workflows.
- Open Access
- 4 March 2026
- Published by Springer Nature in Research Square
Interspecies Interactions Drive Community-Level Selection in Microbial Coalescence
It has long been debated in ecology whether communities behave as cohesive units or as loose collections of independent species. Here, we study this question in the context of community coalescence, the mixing of previously isolated communities, using bacterial microcosm experiments combined with ecological modeling. Our results demonstrate that interspecies interaction strength determines whether communities or species are the units of selection during coalescence. When interactions are moderate to strong, one parental community consistently outcompetes the other, indicating community-level selection. In contrast, under weak interactions, species fates are uncorrelated and the two communities contribute equally to the coalesced outcome, indicating the absence of community-level selection. These patterns extend to communities derived from natural samples with greater taxonomic diversity and richness. Furthermore, we identify two distinct regimes underlying community-level selection in experiments with different media conditions: an emergent regime in which collective dynamics shape outcomes that cannot be predicted from species traits alone, and a top-down regime where dominant species determine the winning community. Together, these results reconcile conflicting observations on community-level selection during community coalescence by demonstrating that communities behave as cohesive units only when interactions are sufficiently strong.
- 4 March 2026
- Published by Springer Nature in Research Square
Open AccessInterspecies Interactions Drive Community-Level Selection in Microbial Coalescence
It has long been debated in ecology whether communities behave as cohesive units or as loose collections of independent species. Here, we study this question in the context of community coalescence, the mixing of previously isolated communities, using bacterial microcosm experiments combined with ecological modeling. Our results demonstrate that interspecies interaction strength determines whether communities or species are the units of selection during coalescence. When interactions are moderate to strong, one parental community consistently outcompetes the other, indicating community-level selection. In contrast, under weak interactions, species fates are uncorrelated and the two communities contribute equally to the coalesced outcome, indicating the absence of community-level selection. These patterns extend to communities derived from natural samples with greater taxonomic diversity and richness. Furthermore, we identify two distinct regimes underlying community-level selection in experiments with different media conditions: an emergent regime in which collective dynamics shape outcomes that cannot be predicted from species traits alone, and a top-down regime where dominant species determine the winning community. Together, these results reconcile conflicting observations on community-level selection during community coalescence by demonstrating that communities behave as cohesive units only when interactions are sufficiently strong.
- Open Access
Interspecies Interactions Drive Community-Level Selection in Microbial Coalescence
It has long been debated in ecology whether communities behave as cohesive units or as loose collections of independent species. Here, we study this question in the context of community coalescence, the mixing of previously isolated communities, using bacterial microcosm experiments combined with ecological modeling. Our results demonstrate that interspecies interaction strength determines whether communities or species are the units of selection during coalescence. When interactions are moderate to strong, one parental community consistently outcompetes the other, indicating community-level selection. In contrast, under weak interactions, species fates are uncorrelated and the two communities contribute equally to the coalesced outcome, indicating the absence of community-level selection. These patterns extend to communities derived from natural samples with greater taxonomic diversity and richness. Furthermore, we identify two distinct regimes underlying community-level selection in experiments with different media conditions: an emergent regime in which collective dynamics shape outcomes that cannot be predicted from species traits alone, and a top-down regime where dominant species determine the winning community. Together, these results reconcile conflicting observations on community-level selection during community coalescence by demonstrating that communities behave as cohesive units only when interactions are sufficiently strong.
Open AccessInterspecies Interactions Drive Community-Level Selection in Microbial Coalescence
It has long been debated in ecology whether communities behave as cohesive units or as loose collections of independent species. Here, we study this question in the context of community coalescence, the mixing of previously isolated communities, using bacterial microcosm experiments combined with ecological modeling. Our results demonstrate that interspecies interaction strength determines whether communities or species are the units of selection during coalescence. When interactions are moderate to strong, one parental community consistently outcompetes the other, indicating community-level selection. In contrast, under weak interactions, species fates are uncorrelated and the two communities contribute equally to the coalesced outcome, indicating the absence of community-level selection. These patterns extend to communities derived from natural samples with greater taxonomic diversity and richness. Furthermore, we identify two distinct regimes underlying community-level selection in experiments with different media conditions: an emergent regime in which collective dynamics shape outcomes that cannot be predicted from species traits alone, and a top-down regime where dominant species determine the winning community. Together, these results reconcile conflicting observations on community-level selection during community coalescence by demonstrating that communities behave as cohesive units only when interactions are sufficiently strong.
- Open Access
- 22 November 2024
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 12 (1) , e59396
A Multivariable Prediction Model for Mild Cognitive Impairment and Dementia: Algorithm Development and Validation
Background: Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic and social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise in managing complex data for MCI and dementia prediction. Objective: This study assessed the predictive accuracy of ML models in identifying the onset of MCI and dementia using the Korean Longitudinal Study of Aging (KLoSA) dataset. Methods: This study used data from the KLoSA, a comprehensive biennial survey that tracks the demographic, health, and socioeconomic aspects of middle-aged and older Korean adults from 2018 to 2020. Among the 6171 initial households, 4975 eligible older adult participants aged 60 years or older were selected after excluding individuals based on age and missing data. The identification of MCI and dementia relied on self-reported diagnoses, with sociodemographic and health-related variables serving as key covariates. The dataset was categorized into training and test sets to predict MCI and dementia by using multiple models, including logistic regression, light gradient-boosting machine, XGBoost (extreme gradient boosting), CatBoost, random forest, gradient boosting, AdaBoost, support vector classifier, and k-nearest neighbors, and the training and test sets were used to evaluate predictive performance. The performance was assessed using the area under the receiver operating characteristic curve (AUC). Class imbalances were addressed via weights. Shapley additive explanation values were used to determine the contribution of each feature to the prediction rate. Results: Among the 4975 participants, the best model for predicting MCI onset was random forest, with a median AUC of 0.6729 (IQR 0.3883-0.8152), followed by k-nearest neighbors with a median AUC of 0.5576 (IQR 0.4555-0.6761) and support vector classifier with a median AUC of 0.5067 (IQR 0.3755-0.6389). For dementia onset prediction, the best model was XGBoost, achieving a median AUC of 0.8185 (IQR 0.8085-0.8285), closely followed by light gradient-boosting machine with a median AUC of 0.8069 (IQR 0.7969-0.8169) and AdaBoost with a median AUC of 0.8007 (IQR 0.7907-0.8107). The Shapley values highlighted pain in everyday life, being widowed, living alone, exercising, and living with a partner as the strongest predictors of MCI. For dementia, the most predictive features were other contributing factors, education at the high school level, education at the middle school level, exercising, and monthly social engagement. Conclusions: ML algorithms, especially XGBoost, exhibited the potential for predicting MCI onset using KLoSA data. However, no model has demonstrated robust accuracy in predicting MCI and dementia. Sociodemographic and health-related factors are crucial for initiating cognitive conditions, emphasizing the need for multifaceted predictive models for early identification and intervention. These findings underscore the potential and limitations of ML in predicting cognitive impairment in community-dwelling older adults.
- 22 November 2024
- Published by JMIR Publications Inc. in JMIR Medical Informatics
- Vol. 12 (1) , e59396
Open AccessA Multivariable Prediction Model for Mild Cognitive Impairment and Dementia: Algorithm Development and Validation
Background: Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic and social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise in managing complex data for MCI and dementia prediction. Objective: This study assessed the predictive accuracy of ML models in identifying the onset of MCI and dementia using the Korean Longitudinal Study of Aging (KLoSA) dataset. Methods: This study used data from the KLoSA, a comprehensive biennial survey that tracks the demographic, health, and socioeconomic aspects of middle-aged and older Korean adults from 2018 to 2020. Among the 6171 initial households, 4975 eligible older adult participants aged 60 years or older were selected after excluding individuals based on age and missing data. The identification of MCI and dementia relied on self-reported diagnoses, with sociodemographic and health-related variables serving as key covariates. The dataset was categorized into training and test sets to predict MCI and dementia by using multiple models, including logistic regression, light gradient-boosting machine, XGBoost (extreme gradient boosting), CatBoost, random forest, gradient boosting, AdaBoost, support vector classifier, and k-nearest neighbors, and the training and test sets were used to evaluate predictive performance. The performance was assessed using the area under the receiver operating characteristic curve (AUC). Class imbalances were addressed via weights. Shapley additive explanation values were used to determine the contribution of each feature to the prediction rate. Results: Among the 4975 participants, the best model for predicting MCI onset was random forest, with a median AUC of 0.6729 (IQR 0.3883-0.8152), followed by k-nearest neighbors with a median AUC of 0.5576 (IQR 0.4555-0.6761) and support vector classifier with a median AUC of 0.5067 (IQR 0.3755-0.6389). For dementia onset prediction, the best model was XGBoost, achieving a median AUC of 0.8185 (IQR 0.8085-0.8285), closely followed by light gradient-boosting machine with a median AUC of 0.8069 (IQR 0.7969-0.8169) and AdaBoost with a median AUC of 0.8007 (IQR 0.7907-0.8107). The Shapley values highlighted pain in everyday life, being widowed, living alone, exercising, and living with a partner as the strongest predictors of MCI. For dementia, the most predictive features were other contributing factors, education at the high school level, education at the middle school level, exercising, and monthly social engagement. Conclusions: ML algorithms, especially XGBoost, exhibited the potential for predicting MCI onset using KLoSA data. However, no model has demonstrated robust accuracy in predicting MCI and dementia. Sociodemographic and health-related factors are crucial for initiating cognitive conditions, emphasizing the need for multifaceted predictive models for early identification and intervention. These findings underscore the potential and limitations of ML in predicting cognitive impairment in community-dwelling older adults.
- Open Access
- 11 April 2024
- Published by JMIR Publications Inc. in JMIR Preprints
A Multivariable Prediction Model for Mild Cognitive Impairment and Dementia: Algorithm Development and Validation (Preprint)
BACKGROUND Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic and social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise in managing complex data for MCI and dementia prediction. OBJECTIVE This study assessed the predictive accuracy of ML models in identifying the onset of MCI and dementia using the Korean Longitudinal Study of Aging (KLoSA) dataset. METHODS This study used data from the KLoSA, a comprehensive biennial survey that tracks the demographic, health, and socioeconomic aspects of middle-aged and older Korean adults from 2018 to 2020. Among the 6171 initial households, 4975 eligible older adult participants aged 60 years or older were selected after excluding individuals based on age and missing data. The identification of MCI and dementia relied on self-reported diagnoses, with sociodemographic and health-related variables serving as key covariates. The dataset was categorized into training and test sets to predict MCI and dementia by using multiple models, including logistic regression, light gradient-boosting machine, XGBoost (extreme gradient boosting), CatBoost, random forest, gradient boosting, AdaBoost, support vector classifier, and k-nearest neighbors, and the training and test sets were used to evaluate predictive performance. The performance was assessed using the area under the receiver operating characteristic curve (AUC). Class imbalances were addressed via weights. Shapley additive explanation values were used to determine the contribution of each feature to the prediction rate. RESULTS Among the 4975 participants, the best model for predicting MCI onset was random forest, with a median AUC of 0.6729 (IQR 0.3883-0.8152), followed by k-nearest neighbors with a median AUC of 0.5576 (IQR 0.4555-0.6761) and support vector classifier with a median AUC of 0.5067 (IQR 0.3755-0.6389). For dementia onset prediction, the best model was XGBoost, achieving a median AUC of 0.8185 (IQR 0.8085-0.8285), closely followed by light gradient-boosting machine with a median AUC of 0.8069 (IQR 0.7969-0.8169) and AdaBoost with a median AUC of 0.8007 (IQR 0.7907-0.8107). The Shapley values highlighted pain in everyday life, being widowed, living alone, exercising, and living with a partner as the strongest predictors of MCI. For dementia, the most predictive features were other contributing factors, education at the high school level, education at the middle school level, exercising, and monthly social engagement. CONCLUSIONS ML algorithms, especially XGBoost, exhibited the potential for predicting MCI onset using KLoSA data. However, no model has demonstrated robust accuracy in predicting MCI and dementia. Sociodemographic and health-related factors are crucial for initiating cognitive conditions, emphasizing the need for multifaceted predictive models for early identification and intervention. These findings underscore the potential and limitations of ML in predicting cognitive impairment in community-dwelling older adults.
- 11 April 2024
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessA Multivariable Prediction Model for Mild Cognitive Impairment and Dementia: Algorithm Development and Validation (Preprint)
BACKGROUND Mild cognitive impairment (MCI) poses significant challenges in early diagnosis and timely intervention. Underdiagnosis, coupled with the economic and social burden of dementia, necessitates more precise detection methods. Machine learning (ML) algorithms show promise in managing complex data for MCI and dementia prediction. OBJECTIVE This study assessed the predictive accuracy of ML models in identifying the onset of MCI and dementia using the Korean Longitudinal Study of Aging (KLoSA) dataset. METHODS This study used data from the KLoSA, a comprehensive biennial survey that tracks the demographic, health, and socioeconomic aspects of middle-aged and older Korean adults from 2018 to 2020. Among the 6171 initial households, 4975 eligible older adult participants aged 60 years or older were selected after excluding individuals based on age and missing data. The identification of MCI and dementia relied on self-reported diagnoses, with sociodemographic and health-related variables serving as key covariates. The dataset was categorized into training and test sets to predict MCI and dementia by using multiple models, including logistic regression, light gradient-boosting machine, XGBoost (extreme gradient boosting), CatBoost, random forest, gradient boosting, AdaBoost, support vector classifier, and k-nearest neighbors, and the training and test sets were used to evaluate predictive performance. The performance was assessed using the area under the receiver operating characteristic curve (AUC). Class imbalances were addressed via weights. Shapley additive explanation values were used to determine the contribution of each feature to the prediction rate. RESULTS Among the 4975 participants, the best model for predicting MCI onset was random forest, with a median AUC of 0.6729 (IQR 0.3883-0.8152), followed by k-nearest neighbors with a median AUC of 0.5576 (IQR 0.4555-0.6761) and support vector classifier with a median AUC of 0.5067 (IQR 0.3755-0.6389). For dementia onset prediction, the best model was XGBoost, achieving a median AUC of 0.8185 (IQR 0.8085-0.8285), closely followed by light gradient-boosting machine with a median AUC of 0.8069 (IQR 0.7969-0.8169) and AdaBoost with a median AUC of 0.8007 (IQR 0.7907-0.8107). The Shapley values highlighted pain in everyday life, being widowed, living alone, exercising, and living with a partner as the strongest predictors of MCI. For dementia, the most predictive features were other contributing factors, education at the high school level, education at the middle school level, exercising, and monthly social engagement. CONCLUSIONS ML algorithms, especially XGBoost, exhibited the potential for predicting MCI onset using KLoSA data. However, no model has demonstrated robust accuracy in predicting MCI and dementia. Sociodemographic and health-related factors are crucial for initiating cognitive conditions, emphasizing the need for multifaceted predictive models for early identification and intervention. These findings underscore the potential and limitations of ML in predicting cognitive impairment in community-dwelling older adults.
- Open Access
- 18 July 2023
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 25, e45041
Predicting Fetal Alcohol Spectrum Disorders Using Machine Learning Techniques: Multisite Retrospective Cohort Study
Background: Fetal alcohol syndrome (FAS) is a lifelong developmental disability that occurs among individuals with prenatal alcohol exposure (PAE). With improved prediction models, FAS can be diagnosed or treated early, if not completely prevented. Objective: In this study, we sought to compare different machine learning algorithms and their FAS predictive performance among women who consumed alcohol during pregnancy. We also aimed to identify which variables (eg, timing of exposure to alcohol during pregnancy and type of alcohol consumed) were most influential in generating an accurate model. Methods: Data from the collaborative initiative on fetal alcohol spectrum disorders from 2007 to 2017 were used to gather information about 595 women who consumed alcohol during pregnancy at 5 hospital sites around the United States. To obtain information about PAE, questionnaires or in-person interviews, as well as reviews of medical, legal, or social service records were used to gather information about alcohol consumption. Four different machine learning algorithms (logistic regression, XGBoost, light gradient-boosting machine, and CatBoost) were trained to predict the prevalence of FAS at birth, and model performance was measured by analyzing the area under the receiver operating characteristics curve (AUROC). Of the total cases, 80% were randomly selected for training, while 20% remained as test data sets for predicting FAS. Feature importance was also analyzed using Shapley values for the best-performing algorithm. Results: Overall, there were 20 cases of FAS within a total population of 595 individuals with PAE. Most of the drinking occurred in the first trimester only (n=491) or throughout all 3 trimesters (n=95); however, there were also reports of drinking in the first and second trimesters only (n=8), and 1 case of drinking in the third trimester only (n=1). The CatBoost method delivered the best performance in terms of AUROC (0.92) and area under the precision-recall curve (AUPRC 0.51), followed by the logistic regression method (AUROC 0.90; AUPRC 0.59), the light gradient-boosting machine (AUROC 0.89; AUPRC 0.52), and XGBoost (AUROC 0.86; AURPC 0.45). Shapley values in the CatBoost model revealed that 12 variables were considered important in FAS prediction, with drinking throughout all 3 trimesters of pregnancy, maternal age, race, and type of alcoholic beverage consumed (eg, beer, wine, or liquor) scoring highly in overall feature importance. For most predictive measures, the best performance was obtained by the CatBoost algorithm, with an AUROC of 0.92, precision of 0.50, specificity of 0.29, F1 score of 0.29, and accuracy of 0.96. Conclusions: Machine learning algorithms were able to identify FAS risk with a prediction performance higher than that of previous models among pregnant drinkers. For small training sets, which are common with FAS, boosting mechanisms like CatBoost may help alleviate certain problems associated with data imbalances and difficulties in optimization or generalization.
- 18 July 2023
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 25, e45041
Open AccessPredicting Fetal Alcohol Spectrum Disorders Using Machine Learning Techniques: Multisite Retrospective Cohort Study
Background: Fetal alcohol syndrome (FAS) is a lifelong developmental disability that occurs among individuals with prenatal alcohol exposure (PAE). With improved prediction models, FAS can be diagnosed or treated early, if not completely prevented. Objective: In this study, we sought to compare different machine learning algorithms and their FAS predictive performance among women who consumed alcohol during pregnancy. We also aimed to identify which variables (eg, timing of exposure to alcohol during pregnancy and type of alcohol consumed) were most influential in generating an accurate model. Methods: Data from the collaborative initiative on fetal alcohol spectrum disorders from 2007 to 2017 were used to gather information about 595 women who consumed alcohol during pregnancy at 5 hospital sites around the United States. To obtain information about PAE, questionnaires or in-person interviews, as well as reviews of medical, legal, or social service records were used to gather information about alcohol consumption. Four different machine learning algorithms (logistic regression, XGBoost, light gradient-boosting machine, and CatBoost) were trained to predict the prevalence of FAS at birth, and model performance was measured by analyzing the area under the receiver operating characteristics curve (AUROC). Of the total cases, 80% were randomly selected for training, while 20% remained as test data sets for predicting FAS. Feature importance was also analyzed using Shapley values for the best-performing algorithm. Results: Overall, there were 20 cases of FAS within a total population of 595 individuals with PAE. Most of the drinking occurred in the first trimester only (n=491) or throughout all 3 trimesters (n=95); however, there were also reports of drinking in the first and second trimesters only (n=8), and 1 case of drinking in the third trimester only (n=1). The CatBoost method delivered the best performance in terms of AUROC (0.92) and area under the precision-recall curve (AUPRC 0.51), followed by the logistic regression method (AUROC 0.90; AUPRC 0.59), the light gradient-boosting machine (AUROC 0.89; AUPRC 0.52), and XGBoost (AUROC 0.86; AURPC 0.45). Shapley values in the CatBoost model revealed that 12 variables were considered important in FAS prediction, with drinking throughout all 3 trimesters of pregnancy, maternal age, race, and type of alcoholic beverage consumed (eg, beer, wine, or liquor) scoring highly in overall feature importance. For most predictive measures, the best performance was obtained by the CatBoost algorithm, with an AUROC of 0.92, precision of 0.50, specificity of 0.29, F1 score of 0.29, and accuracy of 0.96. Conclusions: Machine learning algorithms were able to identify FAS risk with a prediction performance higher than that of previous models among pregnant drinkers. For small training sets, which are common with FAS, boosting mechanisms like CatBoost may help alleviate certain problems associated with data imbalances and difficulties in optimization or generalization.
- Open Access
- 30 May 2023
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 12
Noncovalent antibody catenation on a target surface greatly increases the antigen-binding avidity
Immunoglobulin G (IgG) antibodies are widely used for diagnosis and therapy. Given the unique dimeric structure of IgG, we hypothesized that, by genetically fusing a homodimeric protein (catenator) to the C-terminus of IgG, reversible catenation of antibody molecules could be induced on a surface where target antigen molecules are abundant, and that it could be an effective way to greatly enhance the antigen-binding avidity. A thermodynamic simulation showed that quite low homodimerization affinity of a catenator, e.g. dissociation constant of 100 μM, can enhance nanomolar antigen-binding avidity to a picomolar level, and that the fold enhancement sharply depends on the density of the antigen. In a proof-of-concept experiment where antigen molecules are immobilized on a biosensor tip, the C-terminal fusion of a pair of weakly homodimerizing proteins to three different antibodies enhanced the antigen-binding avidity by at least 110 or 304 folds from the intrinsic binding avidity. Compared with the mother antibody, Obinutuzumab(Y101L) which targets CD20, the same antibody with fused catenators exhibited significantly enhanced binding to SU-DHL5 cells. Together, the homodimerization-induced antibody catenation would be a new powerful approach to improve antibody applications, including the detection of scarce biomarkers and targeted anticancer therapies.
- 30 May 2023
- Published by eLife Sciences Publications, Ltd in eLife
- Vol. 12
Open AccessNoncovalent antibody catenation on a target surface greatly increases the antigen-binding avidity
Immunoglobulin G (IgG) antibodies are widely used for diagnosis and therapy. Given the unique dimeric structure of IgG, we hypothesized that, by genetically fusing a homodimeric protein (catenator) to the C-terminus of IgG, reversible catenation of antibody molecules could be induced on a surface where target antigen molecules are abundant, and that it could be an effective way to greatly enhance the antigen-binding avidity. A thermodynamic simulation showed that quite low homodimerization affinity of a catenator, e.g. dissociation constant of 100 μM, can enhance nanomolar antigen-binding avidity to a picomolar level, and that the fold enhancement sharply depends on the density of the antigen. In a proof-of-concept experiment where antigen molecules are immobilized on a biosensor tip, the C-terminal fusion of a pair of weakly homodimerizing proteins to three different antibodies enhanced the antigen-binding avidity by at least 110 or 304 folds from the intrinsic binding avidity. Compared with the mother antibody, Obinutuzumab(Y101L) which targets CD20, the same antibody with fused catenators exhibited significantly enhanced binding to SU-DHL5 cells. Together, the homodimerization-induced antibody catenation would be a new powerful approach to improve antibody applications, including the detection of scarce biomarkers and targeted anticancer therapies.
- Open Access
- 13 December 2022
- Published by JMIR Publications Inc. in JMIR Preprints
Predicting Fetal Alcohol Spectrum Disorders Using Machine Learning Techniques: Multisite Retrospective Cohort Study (Preprint)
BACKGROUND Fetal alcohol syndrome (FAS) is a lifelong developmental disability that occurs among individuals with prenatal alcohol exposure (PAE). With improved prediction models, FAS can be diagnosed or treated early, if not completely prevented. OBJECTIVE In this study, we sought to compare different machine learning algorithms and their FAS predictive performance among women who consumed alcohol during pregnancy. We also aimed to identify which variables (eg, timing of exposure to alcohol during pregnancy and type of alcohol consumed) were most influential in generating an accurate model. METHODS Data from the collaborative initiative on fetal alcohol spectrum disorders from 2007 to 2017 were used to gather information about 595 women who consumed alcohol during pregnancy at 5 hospital sites around the United States. To obtain information about PAE, questionnaires or in-person interviews, as well as reviews of medical, legal, or social service records were used to gather information about alcohol consumption. Four different machine learning algorithms (logistic regression, XGBoost, light gradient-boosting machine, and CatBoost) were trained to predict the prevalence of FAS at birth, and model performance was measured by analyzing the area under the receiver operating characteristics curve (AUROC). Of the total cases, 80% were randomly selected for training, while 20% remained as test data sets for predicting FAS. Feature importance was also analyzed using Shapley values for the best-performing algorithm. RESULTS Overall, there were 20 cases of FAS within a total population of 595 individuals with PAE. Most of the drinking occurred in the first trimester only (n=491) or throughout all 3 trimesters (n=95); however, there were also reports of drinking in the first and second trimesters only (n=8), and 1 case of drinking in the third trimester only (n=1). The CatBoost method delivered the best performance in terms of AUROC (0.92) and area under the precision-recall curve (AUPRC 0.51), followed by the logistic regression method (AUROC 0.90; AUPRC 0.59), the light gradient-boosting machine (AUROC 0.89; AUPRC 0.52), and XGBoost (AUROC 0.86; AURPC 0.45). Shapley values in the CatBoost model revealed that 12 variables were considered important in FAS prediction, with drinking throughout all 3 trimesters of pregnancy, maternal age, race, and type of alcoholic beverage consumed (eg, beer, wine, or liquor) scoring highly in overall feature importance. For most predictive measures, the best performance was obtained by the CatBoost algorithm, with an AUROC of 0.92, precision of 0.50, specificity of 0.29, F1 score of 0.29, and accuracy of 0.96. CONCLUSIONS Machine learning algorithms were able to identify FAS risk with a prediction performance higher than that of previous models among pregnant drinkers. For small training sets, which are common with FAS, boosting mechanisms like CatBoost may help alleviate certain problems associated with data imbalances and difficulties in optimization or generalization.
- 13 December 2022
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessPredicting Fetal Alcohol Spectrum Disorders Using Machine Learning Techniques: Multisite Retrospective Cohort Study (Preprint)
BACKGROUND Fetal alcohol syndrome (FAS) is a lifelong developmental disability that occurs among individuals with prenatal alcohol exposure (PAE). With improved prediction models, FAS can be diagnosed or treated early, if not completely prevented. OBJECTIVE In this study, we sought to compare different machine learning algorithms and their FAS predictive performance among women who consumed alcohol during pregnancy. We also aimed to identify which variables (eg, timing of exposure to alcohol during pregnancy and type of alcohol consumed) were most influential in generating an accurate model. METHODS Data from the collaborative initiative on fetal alcohol spectrum disorders from 2007 to 2017 were used to gather information about 595 women who consumed alcohol during pregnancy at 5 hospital sites around the United States. To obtain information about PAE, questionnaires or in-person interviews, as well as reviews of medical, legal, or social service records were used to gather information about alcohol consumption. Four different machine learning algorithms (logistic regression, XGBoost, light gradient-boosting machine, and CatBoost) were trained to predict the prevalence of FAS at birth, and model performance was measured by analyzing the area under the receiver operating characteristics curve (AUROC). Of the total cases, 80% were randomly selected for training, while 20% remained as test data sets for predicting FAS. Feature importance was also analyzed using Shapley values for the best-performing algorithm. RESULTS Overall, there were 20 cases of FAS within a total population of 595 individuals with PAE. Most of the drinking occurred in the first trimester only (n=491) or throughout all 3 trimesters (n=95); however, there were also reports of drinking in the first and second trimesters only (n=8), and 1 case of drinking in the third trimester only (n=1). The CatBoost method delivered the best performance in terms of AUROC (0.92) and area under the precision-recall curve (AUPRC 0.51), followed by the logistic regression method (AUROC 0.90; AUPRC 0.59), the light gradient-boosting machine (AUROC 0.89; AUPRC 0.52), and XGBoost (AUROC 0.86; AURPC 0.45). Shapley values in the CatBoost model revealed that 12 variables were considered important in FAS prediction, with drinking throughout all 3 trimesters of pregnancy, maternal age, race, and type of alcoholic beverage consumed (eg, beer, wine, or liquor) scoring highly in overall feature importance. For most predictive measures, the best performance was obtained by the CatBoost algorithm, with an AUROC of 0.92, precision of 0.50, specificity of 0.29, F1 score of 0.29, and accuracy of 0.96. CONCLUSIONS Machine learning algorithms were able to identify FAS risk with a prediction performance higher than that of previous models among pregnant drinkers. For small training sets, which are common with FAS, boosting mechanisms like CatBoost may help alleviate certain problems associated with data imbalances and difficulties in optimization or generalization.
- Open Access
Noncovalent antibody catenation on a target surface drastically increases the antigen-binding avidity
Immunoglobulin G (IgG) antibodies are widely used for diagnosis and therapy. Given the unique dimeric structure of IgG, we hypothesized that, by genetically fusing a homodimeric protein (catenator) to the C-terminus of IgG, reversible catenation of antibody molecules could be induced on a surface where target antigen molecules are abundant, and that it could be an effective way to greatly enhance the antigen-binding avidity. A thermodynamic simulation shows that quite low homodimerization affinity of a catenator, e.g. dissociation constant of 100 μM, can enhance nanomolar antigen-binding avidity to a picomolar level, and that the fold enhancement sharply depends on the density of the antigen. In a proof-of-concept experiment where antigen molecules are immobilized on a biosensor tip, C-terminal fusion of a weakly homodimerizing protein to two different antibodies enhanced the antigen-binding avidity by at least 210 to 5,120 folds from the intrinsic binding avidity. Thus, the homodimerization-induced antibody catenation would be a simple, powerful and general approach to improve many antibody applications, including the detection of scarce biomarkers and targeted anticancer therapies.
Open AccessNoncovalent antibody catenation on a target surface drastically increases the antigen-binding avidity
Immunoglobulin G (IgG) antibodies are widely used for diagnosis and therapy. Given the unique dimeric structure of IgG, we hypothesized that, by genetically fusing a homodimeric protein (catenator) to the C-terminus of IgG, reversible catenation of antibody molecules could be induced on a surface where target antigen molecules are abundant, and that it could be an effective way to greatly enhance the antigen-binding avidity. A thermodynamic simulation shows that quite low homodimerization affinity of a catenator, e.g. dissociation constant of 100 μM, can enhance nanomolar antigen-binding avidity to a picomolar level, and that the fold enhancement sharply depends on the density of the antigen. In a proof-of-concept experiment where antigen molecules are immobilized on a biosensor tip, C-terminal fusion of a weakly homodimerizing protein to two different antibodies enhanced the antigen-binding avidity by at least 210 to 5,120 folds from the intrinsic binding avidity. Thus, the homodimerization-induced antibody catenation would be a simple, powerful and general approach to improve many antibody applications, including the detection of scarce biomarkers and targeted anticancer therapies.
- 11 March 2020
- Published by SPIE-Intl Soc Optical Eng
- Vol. 11249, 112490Y-33
Deep learning framework enables 3D label-free tracking of immunological synapse using optical diffraction tomography (Conference Presentation)
Rapid, label-free, volumetric, and automated assessment in microscopy is necessary to assess the dynamic interactions between lymphocytes and their targets through the immunological synapse (IS) and the relevant immunological functions. However, attempts to realize the automatic tracking of IS dynamics have been stymied by the limitations of imaging techniques and computational analysis methods. Here, we demonstrate the automatic three-dimensional IS tracking by combining optical diffraction tomography and deep-learning-based segmentation. The proposed approach enables quantitative spatiotemporal analyses of IS regarding morphological and biochemical parameters related to its protein densities, offering a novel complementary method to fluorescence microscopy for studies in immunology.
- 11 March 2020
- Published by SPIE-Intl Soc Optical Eng
- Vol. 11249, 112490Y-33
Deep learning framework enables 3D label-free tracking of immunological synapse using optical diffraction tomography (Conference Presentation)
Rapid, label-free, volumetric, and automated assessment in microscopy is necessary to assess the dynamic interactions between lymphocytes and their targets through the immunological synapse (IS) and the relevant immunological functions. However, attempts to realize the automatic tracking of IS dynamics have been stymied by the limitations of imaging techniques and computational analysis methods. Here, we demonstrate the automatic three-dimensional IS tracking by combining optical diffraction tomography and deep-learning-based segmentation. The proposed approach enables quantitative spatiotemporal analyses of IS regarding morphological and biochemical parameters related to its protein densities, offering a novel complementary method to fluorescence microscopy for studies in immunology.