Joseph D. Romano
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
Joseph D. Romano has a total of 94 co-authors. Here are the top co-authors by number of shared publications:
- Jason H. Moore
Department of Computational Biomedicine, Cedars-Sinai Medical Center, West Hollywood, USA
8 shared publications - Marylyn DeRiggi Ritchie
Department of Genetics, University of Pennsylvania Perelman School of Medicine, Philadelphia, USA
6 shared publications - Nicholas P. Tatonetti
Department of Computational Biomedicine, Cedars-Sinai Medical Center, Los Angeles, USA
5 shared publications - Rachit Kumar
University of Pennsylvania Perelman School of Medicine
5 shared publications Ananya Rajagopalan
Graduate Group in Genomics and Computational Biology, University of Pennsylvania, Philadelphia, Pennsylvania, USA
2 shared publicationsAndre Luis Garao Rico
Department of Genetics, Philadelphia, PA, USA
2 shared publications
Latest Publications
- Open Access
- 6 November 2026
- Published by American Chemical Society (ACS) in Environmental Health Perspectives
Per- and Polyfluoroalkyl Substances, Hepatic Lipid Metabolism Genes, and Hepatocellular Carcinoma: Evidence of Gene–Environment Interactions in a Multiethnic Cohort
Per- and polyfluoroalkyl substances (PFAS), synthetic pollutants that impact hepatic lipid metabolism, may increase the risk of hepatocellular carcinoma (HCC). We examined associations of prediagnostic blood PFAS levels with HCC risk and examined whether genetic susceptibility to increased hepatic lipid accumulation modified these associations. We analyzed 223 incident HCC cases and 223 matched controls from the Multiethnic Cohort. Concentrations of 12 PFAS were measured in prediagnostic plasma samples. Conditional logistic regression, adjusting for covariates, assessed PFAS–HCC associations. Effect modification was evaluated using a dichotomized polygenic risk score (PRS) for metabolic dysfunction-associated steatotic liver disease (MASLD). Each 2-fold increase in PFHpA and PFNA, respectively, was associated with 1.35 (95% CI: 1.02–1.78; p = 0.035) and 1.31 (1.00–1.72; p = 0.047) times higher HCC odds, respectively; both also exhibited significant dose–response relationships. The association between PFHpA and HCC was stronger in individuals with a high MASLD PRS. Each 2-fold increase in PFHpA was associated with a 1.47 (95% CI: 1.06–2.04; p = 0.021) times higher odds of HCC in individuals with high MASLD genetic risk, but it was not associated with HCC among those with low MASLD genetic risk (odds ratio: 0.98, 95% CI: 0.68–1.42). Although the multiplicative interaction with continuous PFAS did not reach statistical significance (p = 0.088), analyses with categorical PFHpA exposure provided evidence of a positive additive interaction. Prediagnostic concentrations of both short- and long-chain PFAS species were associated with increased HCC risk. Findings for PFHpA suggest that genetic susceptibility to hepatic lipid accumulation may modify PFAS-associated HCC risk, supporting a potential link between environmental exposures, hepatic lipid metabolism, and genetic susceptibility to HCC.
- 6 November 2026
- Published by American Chemical Society (ACS) in Environmental Health Perspectives
Open AccessPer- and Polyfluoroalkyl Substances, Hepatic Lipid Metabolism Genes, and Hepatocellular Carcinoma: Evidence of Gene–Environment Interactions in a Multiethnic Cohort
Per- and polyfluoroalkyl substances (PFAS), synthetic pollutants that impact hepatic lipid metabolism, may increase the risk of hepatocellular carcinoma (HCC). We examined associations of prediagnostic blood PFAS levels with HCC risk and examined whether genetic susceptibility to increased hepatic lipid accumulation modified these associations. We analyzed 223 incident HCC cases and 223 matched controls from the Multiethnic Cohort. Concentrations of 12 PFAS were measured in prediagnostic plasma samples. Conditional logistic regression, adjusting for covariates, assessed PFAS–HCC associations. Effect modification was evaluated using a dichotomized polygenic risk score (PRS) for metabolic dysfunction-associated steatotic liver disease (MASLD). Each 2-fold increase in PFHpA and PFNA, respectively, was associated with 1.35 (95% CI: 1.02–1.78; p = 0.035) and 1.31 (1.00–1.72; p = 0.047) times higher HCC odds, respectively; both also exhibited significant dose–response relationships. The association between PFHpA and HCC was stronger in individuals with a high MASLD PRS. Each 2-fold increase in PFHpA was associated with a 1.47 (95% CI: 1.06–2.04; p = 0.021) times higher odds of HCC in individuals with high MASLD genetic risk, but it was not associated with HCC among those with low MASLD genetic risk (odds ratio: 0.98, 95% CI: 0.68–1.42). Although the multiplicative interaction with continuous PFAS did not reach statistical significance (p = 0.088), analyses with categorical PFHpA exposure provided evidence of a positive additive interaction. Prediagnostic concentrations of both short- and long-chain PFAS species were associated with increased HCC risk. Findings for PFHpA suggest that genetic susceptibility to hepatic lipid accumulation may modify PFAS-associated HCC risk, supporting a potential link between environmental exposures, hepatic lipid metabolism, and genetic susceptibility to HCC.
- Open Access
The relative importance of neighborhood environment features in explaining preeclampsia risk using machine learning
Introduction: Disentangling the role of the neighborhood environment in preeclampsia pathogenesis is crucial for addressing social and structural determinants of pregnancy health. Building on epidemiologic studies demonstrating environmental associations with preeclampsia, we used a machine learning approach to determine the relative importance of multiple simultaneous neighborhood features to facilitate prioritization of community pregnancy health initiatives. Methods: We linked 26 features from the neighborhood environment, encompassing social vulnerability, built environment, physical environment, and health vulnerability features, to geocoded residential addresses of participants selected for a matched, nested case‐control study from two Philadelphia hospitals. We modeled individual associations of neighborhood features with preeclampsia using conditional logistic regression models. We then built XGBoost models trained on the neighborhood features predicting preeclampsia and applied explainable artificial intelligence (XAI) to disentangle the relative importance of the features associated with preeclampsia. Results: Among 18,754 participants (4689 preeclampsia cases and 14,065 controls), we observed significant associations of neighborhood health and social vulnerability features with preeclampsia. From the XGBoost models, three neighborhood health vulnerability features—prevalence of obesity, prevalence of high blood pressure, and prevalence of short sleep duration among adults—were the most important neighborhood features in predicting preeclampsia. Conclusion: Our findings that neighborhood features vary with respect to their relative importance in predicting preeclampsia demonstrate the value of using XAI to contribute new insights into the vulnerability of pregnant individuals to specific neighborhood environmental features and to inform policy‐making priorities for community‐level pregnancy health interventions. Specifically, the association between neighborhood hypertension prevalence and preeclampsia, suggests that communities with worse cardiovascular health have a higher preeclampsia risk, which warrants further investigation as a potential avenue for preeclampsia prevention.
Open AccessThe relative importance of neighborhood environment features in explaining preeclampsia risk using machine learning
Introduction: Disentangling the role of the neighborhood environment in preeclampsia pathogenesis is crucial for addressing social and structural determinants of pregnancy health. Building on epidemiologic studies demonstrating environmental associations with preeclampsia, we used a machine learning approach to determine the relative importance of multiple simultaneous neighborhood features to facilitate prioritization of community pregnancy health initiatives. Methods: We linked 26 features from the neighborhood environment, encompassing social vulnerability, built environment, physical environment, and health vulnerability features, to geocoded residential addresses of participants selected for a matched, nested case‐control study from two Philadelphia hospitals. We modeled individual associations of neighborhood features with preeclampsia using conditional logistic regression models. We then built XGBoost models trained on the neighborhood features predicting preeclampsia and applied explainable artificial intelligence (XAI) to disentangle the relative importance of the features associated with preeclampsia. Results: Among 18,754 participants (4689 preeclampsia cases and 14,065 controls), we observed significant associations of neighborhood health and social vulnerability features with preeclampsia. From the XGBoost models, three neighborhood health vulnerability features—prevalence of obesity, prevalence of high blood pressure, and prevalence of short sleep duration among adults—were the most important neighborhood features in predicting preeclampsia. Conclusion: Our findings that neighborhood features vary with respect to their relative importance in predicting preeclampsia demonstrate the value of using XAI to contribute new insights into the vulnerability of pregnant individuals to specific neighborhood environmental features and to inform policy‐making priorities for community‐level pregnancy health interventions. Specifically, the association between neighborhood hypertension prevalence and preeclampsia, suggests that communities with worse cardiovascular health have a higher preeclampsia risk, which warrants further investigation as a potential avenue for preeclampsia prevention.
- 8 July 2026
- Published by Springer Nature
- p. 68-72
A Phyto-Ontology for Integrative Therapeutic Discovery
- 8 July 2026
- Published by Springer Nature
- p. 68-72
A Phyto-Ontology for Integrative Therapeutic Discovery
- Open Access
- 28 April 2026
- Published by Annual Reviews in Annual Review of Biomedical Data Science
- Vol. 9 (1) , 239-263
Clinical Knowledge Representation in Data Science
The vast potential of observational healthcare data in biomedical discovery remains largely unrealized because clinical records are fragmented, unstructured, and generated for patient care rather than research. Clinical knowledge representation (KR) helps to bridge this gap by encoding information in standardized, computable formats that preserve meaning and context. This review examines KR across the clinical data life cycle, from the generation of data in healthcare settings to their transformation for secondary use and eventual application in data science. We highlight foundational components such as standardized terminologies, ontologies, and common data models that enable data harmonization and interoperability. We further discuss how these structured representations support multimodal data integration and the development of more accurate, interpretable artificial intelligence models. Adopting a semantics-first approach to KR is essential for transforming fragmented clinical data into reusable, trustworthy knowledge that advances data-driven discovery and improves patient care.
- 28 April 2026
- Published by Annual Reviews in Annual Review of Biomedical Data Science
- Vol. 9 (1) , 239-263
Open AccessClinical Knowledge Representation in Data Science
The vast potential of observational healthcare data in biomedical discovery remains largely unrealized because clinical records are fragmented, unstructured, and generated for patient care rather than research. Clinical knowledge representation (KR) helps to bridge this gap by encoding information in standardized, computable formats that preserve meaning and context. This review examines KR across the clinical data life cycle, from the generation of data in healthcare settings to their transformation for secondary use and eventual application in data science. We highlight foundational components such as standardized terminologies, ontologies, and common data models that enable data harmonization and interoperability. We further discuss how these structured representations support multimodal data integration and the development of more accurate, interpretable artificial intelligence models. Adopting a semantics-first approach to KR is essential for transforming fragmented clinical data into reusable, trustworthy knowledge that advances data-driven discovery and improves patient care.
- Open Access
- 9 January 2026
- Published by Springer Nature in Journal of Cheminformatics
- Vol. 18 (1)
Learnable protein representations in computational biology for predicting drug-target affinity
In this review, we discuss the various different types of learnable protein representations that have been used in computational biology, with a particular
- 9 January 2026
- Published by Springer Nature in Journal of Cheminformatics
- Vol. 18 (1)
Open AccessLearnable protein representations in computational biology for predicting drug-target affinity
In this review, we discuss the various different types of learnable protein representations that have been used in computational biology, with a particular
- 14 December 2025
- Published by World Scientific Pub Co Pte Ltd in Biocomputing 2020
- Vol. 31, 1-11
Session Introduction: AI and Machine Learning in Clinical Medicine Bridging or Separating Model Intelligence and Human Expertise
Artificial Intelligence (AI) technologies continue to expand their role in clinical medicine, with large language models (LLMs) and multimodal systems now applied to communication, imaging, and predictive analytics. Advances in generative and retrieval-augmented methods have improved the accuracy and contextual grounding of clinical summaries, patient messaging, and decision support. At the same time, new benchmarks in imaging, vision, and spontaneous speech have underscored both progress and the persistence of unsolved challenges. Predictive modeling efforts highlight causality, longitudinal trajectories, and informative clinical events, while methodological contributions emphasize uncertainty management, abstention, and interpretable causal structures. Finally, frameworks for evaluation and governance address the crucial gap between laboratory performance and real-world deployment.
- 14 December 2025
- Published by World Scientific Pub Co Pte Ltd in Biocomputing 2020
- Vol. 31, 1-11
Session Introduction: AI and Machine Learning in Clinical Medicine Bridging or Separating Model Intelligence and Human Expertise
Artificial Intelligence (AI) technologies continue to expand their role in clinical medicine, with large language models (LLMs) and multimodal systems now applied to communication, imaging, and predictive analytics. Advances in generative and retrieval-augmented methods have improved the accuracy and contextual grounding of clinical summaries, patient messaging, and decision support. At the same time, new benchmarks in imaging, vision, and spontaneous speech have underscored both progress and the persistence of unsolved challenges. Predictive modeling efforts highlight causality, longitudinal trajectories, and informative clinical events, while methodological contributions emphasize uncertainty management, abstention, and interpretable causal structures. Finally, frameworks for evaluation and governance address the crucial gap between laboratory performance and real-world deployment.
- Open Access
- 14 December 2025
- Published by World Scientific Pub Co Pte Ltd in Biocomputing 2020
- Vol. 31, 830-848
DRIVE-KG: Enhancing variant-phenotype association discovery in understudied complex diseases using heterogeneous knowledge graphs
Multi-omics data are instrumental in obtaining a comprehensive picture of complex biological systems. This is particularly useful for women's health conditions such as endometriosis, which has been historically understudied despite having a high prevalence (around 10% of women of reproductive age). Subsequently, endometriosis has limited genetic characterization: current genome-wide association studies explain only 11% of its 47% total estimated heritability, underscoring the need for integrative approaches. Graph representations provide an intuitive and meaningful way to harmonize biological data, using nodes to represent biological concepts (e.g., genes, single nucleotide polymorphisms, proteins, and phenotypes) and edges to represent their relationships. We present DRIVE-KG (Disease Risk Inference and Variant Exploration Knowledge Graph), which uses a heterogeneous graph representation to integrate data from diverse multi-omics datasets. We trained two distinct models using DRIVE-KG: a link prediction model to suggest associations between SNPs and two pilot phenotypes (endometriosis and obesity), and a graph convolutional network (GCN) for patient-level classification of endometriosis/adenomyosis as a combined phenotype. We conducted patient-level classification using data from 1,441 Penn Medicine BioBank participants with gold standard chart-reviewed endometriosis/adenomyosis status. The link prediction model uncovered 66 high-confidence (model score ≥ 0.95) candidate SNP-endometriosis associations, representing largely distinct genetic signals (R2 < 0.1). These variants were enriched for obesity/body mass index traits (24.2%), lipid metabolism (6%), and depressive disorders (4.5%), showing agreement with emerging hypotheses about endometriosis etiology. In contrast, of the high-confidence, candidate SNP-obesity associations that could be evaluated using LDlink, 38.22% were in high linkage disequilibrium (R2 ≥ 0.8) with known obesity or comorbidity associations. The GCN to classify patient endometriosis/adenomyosis status had an F1 score of 0.752 compared to 0.698 for a genetic risk score. Despite this moderate improvement, we found that the GCN learned meaningful stratification of underlying adenomyosis signal and severe endometriosis grades. Together, these results demonstrate that heterogeneous integration of multi-omics data is valuable for diverse downstream tasks-including discovery and clinical prediction-particularly for understudied diseases where traditional genomic approaches are insufficient.
- 14 December 2025
- Published by World Scientific Pub Co Pte Ltd in Biocomputing 2020
- Vol. 31, 830-848
Open AccessDRIVE-KG: Enhancing variant-phenotype association discovery in understudied complex diseases using heterogeneous knowledge graphs
Multi-omics data are instrumental in obtaining a comprehensive picture of complex biological systems. This is particularly useful for women's health conditions such as endometriosis, which has been historically understudied despite having a high prevalence (around 10% of women of reproductive age). Subsequently, endometriosis has limited genetic characterization: current genome-wide association studies explain only 11% of its 47% total estimated heritability, underscoring the need for integrative approaches. Graph representations provide an intuitive and meaningful way to harmonize biological data, using nodes to represent biological concepts (e.g., genes, single nucleotide polymorphisms, proteins, and phenotypes) and edges to represent their relationships. We present DRIVE-KG (Disease Risk Inference and Variant Exploration Knowledge Graph), which uses a heterogeneous graph representation to integrate data from diverse multi-omics datasets. We trained two distinct models using DRIVE-KG: a link prediction model to suggest associations between SNPs and two pilot phenotypes (endometriosis and obesity), and a graph convolutional network (GCN) for patient-level classification of endometriosis/adenomyosis as a combined phenotype. We conducted patient-level classification using data from 1,441 Penn Medicine BioBank participants with gold standard chart-reviewed endometriosis/adenomyosis status. The link prediction model uncovered 66 high-confidence (model score ≥ 0.95) candidate SNP-endometriosis associations, representing largely distinct genetic signals (R2 < 0.1). These variants were enriched for obesity/body mass index traits (24.2%), lipid metabolism (6%), and depressive disorders (4.5%), showing agreement with emerging hypotheses about endometriosis etiology. In contrast, of the high-confidence, candidate SNP-obesity associations that could be evaluated using LDlink, 38.22% were in high linkage disequilibrium (R2 ≥ 0.8) with known obesity or comorbidity associations. The GCN to classify patient endometriosis/adenomyosis status had an F1 score of 0.752 compared to 0.698 for a genetic risk score. Despite this moderate improvement, we found that the GCN learned meaningful stratification of underlying adenomyosis signal and severe endometriosis grades. Together, these results demonstrate that heterogeneous integration of multi-omics data is valuable for diverse downstream tasks-including discovery and clinical prediction-particularly for understudied diseases where traditional genomic approaches are insufficient.
- Open Access
- 31 August 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (5)
CASTER-DTA: equivariant graph neural networks for predicting drug–target affinity
Accurately determining the binding affinity of a ligand with a protein is important for drug design, development, and screening. With the advent of accessible protein structure prediction methods such as AlphaFold, predicted protein 3D structures are readily available; however, scalable methods for predicting binding affinity currently do not take full advantage of 3D protein information. Here, we present CASTER-DTA (Cross-Attention with Structural Target Equivariant Representations for Drug–Target Affinity), which uses an equivariant graph neural network (GNN) to learn more robust protein representations alongside a standard GNN to learn molecular representations to predict DTA. We augment these representations by incorporating an attention-based mechanism between protein residues and drug atoms to improve interpretability. We show that CASTER-DTA represents a state-of-the-art improvement on multiple benchmarks for predicting DTA, and that it generates novel insights for several related tasks. We then apply CASTER-DTA to create a large resource of the binding affinities of every drug approved by the U.S. Food and Drug Administration (FDA) against every protein in the human proteome and make these predictions freely available for download. We also make available a web server for researchers to apply a pretrained CASTER-DTA model for predicting binding affinities between arbitrary proteins and drugs.
- 31 August 2025
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 26 (5)
Open AccessCASTER-DTA: equivariant graph neural networks for predicting drug–target affinity
Accurately determining the binding affinity of a ligand with a protein is important for drug design, development, and screening. With the advent of accessible protein structure prediction methods such as AlphaFold, predicted protein 3D structures are readily available; however, scalable methods for predicting binding affinity currently do not take full advantage of 3D protein information. Here, we present CASTER-DTA (Cross-Attention with Structural Target Equivariant Representations for Drug–Target Affinity), which uses an equivariant graph neural network (GNN) to learn more robust protein representations alongside a standard GNN to learn molecular representations to predict DTA. We augment these representations by incorporating an attention-based mechanism between protein residues and drug atoms to improve interpretability. We show that CASTER-DTA represents a state-of-the-art improvement on multiple benchmarks for predicting DTA, and that it generates novel insights for several related tasks. We then apply CASTER-DTA to create a large resource of the binding affinities of every drug approved by the U.S. Food and Drug Administration (FDA) against every protein in the human proteome and make these predictions freely available for download. We also make available a web server for researchers to apply a pretrained CASTER-DTA model for predicting binding affinities between arbitrary proteins and drugs.
- Open Access
DRIVE-KG: Enhancing variant-phenotype association discovery in understudied complex diseases using heterogeneous knowledge graphs
Multi-omics data are instrumental in obtaining a comprehensive picture of complex biological systems. This is particularly useful for women’s health conditions, such as endometriosis which has been historically understudied despite having a high prevalence (around 10% of women of reproductive age). Subsequently, endometriosis has limited genetic characterization: current genome-wide association studies explain only 11% of its 47% total estimated heritability. Graph representations provide an intuitive and meaningful way to relate concepts across diverse data sources and address fundamental sparsity and dimensionality challenges with multi-omics data analysis. Here we present DRIVE-KG (Disease Risk Inference and Variant Exploration-Knowledge Graph), which uses a heterogeneous graph representation to integrate biological data from multi-omics datasets: dbSNP, NCBI Human Gene, Omics Pred, GTEx, and Open Targets. We drew directly from the knowledge captured in these data, using nodes to represent genes, single nucleotide polymorphisms, proteins, and phenotypes, and edges to represent relationships between these concepts. We trained two models using DRIVE-KG: a link prediction model to suggest associations between SNPs and two pilot phenotypes (endometriosis and obesity), and a graph convolutional network (GCN) to classify patient-level endometriosis status. We conducted the patient-level classification using data from 1,441 Penn Medicine BioBank participants with gold standard chart-reviewed endometriosis status. The link prediction model uncovered 66 high-confidence (score ≥ 0.95) previously unreported SNP-endometriosis associations. Many of these variants were linked to obesity/body mass index traits (24.2%), lipid metabolism (6%), and depressive disorders (4.5%), showing agreement with emerging hypotheses about endometriosis etiology. In contrast, 11% of the 149 high confidence, candidate SNP-obesity associations (score ≥ 0.9888) were in LD with known obesity associations. The GCN to classify patient endometriosis status had an AUPRC of 0.738 compared to 0.679 for a genetic risk score. Despite this moderate improvement, we found that the GCN learned meaningful stratification of underlying adenomyosis signal and severe grades of endometriosis. We have demonstrated that heterogeneous integration of multi-omics data is valuable for diverse downstream tasks—including discovery and clinical prediction—particularly for understudied diseases where traditional genomic approaches are insufficient.
Open AccessDRIVE-KG: Enhancing variant-phenotype association discovery in understudied complex diseases using heterogeneous knowledge graphs
Multi-omics data are instrumental in obtaining a comprehensive picture of complex biological systems. This is particularly useful for women’s health conditions, such as endometriosis which has been historically understudied despite having a high prevalence (around 10% of women of reproductive age). Subsequently, endometriosis has limited genetic characterization: current genome-wide association studies explain only 11% of its 47% total estimated heritability. Graph representations provide an intuitive and meaningful way to relate concepts across diverse data sources and address fundamental sparsity and dimensionality challenges with multi-omics data analysis. Here we present DRIVE-KG (Disease Risk Inference and Variant Exploration-Knowledge Graph), which uses a heterogeneous graph representation to integrate biological data from multi-omics datasets: dbSNP, NCBI Human Gene, Omics Pred, GTEx, and Open Targets. We drew directly from the knowledge captured in these data, using nodes to represent genes, single nucleotide polymorphisms, proteins, and phenotypes, and edges to represent relationships between these concepts. We trained two models using DRIVE-KG: a link prediction model to suggest associations between SNPs and two pilot phenotypes (endometriosis and obesity), and a graph convolutional network (GCN) to classify patient-level endometriosis status. We conducted the patient-level classification using data from 1,441 Penn Medicine BioBank participants with gold standard chart-reviewed endometriosis status. The link prediction model uncovered 66 high-confidence (score ≥ 0.95) previously unreported SNP-endometriosis associations. Many of these variants were linked to obesity/body mass index traits (24.2%), lipid metabolism (6%), and depressive disorders (4.5%), showing agreement with emerging hypotheses about endometriosis etiology. In contrast, 11% of the 149 high confidence, candidate SNP-obesity associations (score ≥ 0.9888) were in LD with known obesity associations. The GCN to classify patient endometriosis status had an AUPRC of 0.738 compared to 0.679 for a genetic risk score. Despite this moderate improvement, we found that the GCN learned meaningful stratification of underlying adenomyosis signal and severe grades of endometriosis. We have demonstrated that heterogeneous integration of multi-omics data is valuable for diverse downstream tasks—including discovery and clinical prediction—particularly for understudied diseases where traditional genomic approaches are insufficient.