95 results on '"Souparno Ghosh"'
Search Results
2. Image Filtering to Improve Maize Tassel Detection Accuracy Using Machine Learning Algorithms
- Author
-
Eric Rodene, Gayara Demini Fernando, Ved Piyush, Yufeng Ge, James C. Schnable, Souparno Ghosh, and Jinliang Yang
- Subjects
UAV imagery ,high-throughput phenotyping ,machine learning ,convolutional neural network ,object detection ,maize tassel detection ,Chemical technology ,TP1-1185 - Abstract
Unmanned aerial vehicle (UAV)-based imagery has become widely used to collect time-series agronomic data, which are then incorporated into plant breeding programs to enhance crop improvements. To make efficient analysis possible, in this study, by leveraging an aerial photography dataset for a field trial of 233 different inbred lines from the maize diversity panel, we developed machine learning methods for obtaining automated tassel counts at the plot level. We employed both an object-based counting-by-detection (CBD) approach and a density-based counting-by-regression (CBR) approach. Using an image segmentation method that removes most of the pixels not associated with the plant tassels, the results showed a dramatic improvement in the accuracy of object-based (CBD) detection, with the cross-validation prediction accuracy (r2) peaking at 0.7033 on a detector trained with images with a filter threshold of 90. The CBR approach showed the greatest accuracy when using unfiltered images, with a mean absolute error (MAE) of 7.99. However, when using bootstrapping, images filtered at a threshold of 90 showed a slightly better MAE (8.65) than the unfiltered images (8.90). These methods will allow for accurate estimates of flowering-related traits and help to make breeding decisions for crop improvement.
- Published
- 2024
- Full Text
- View/download PDF
3. Representation of features as images with neighborhood dependencies for compatibility with convolutional neural networks
- Author
-
Omid Bazgir, Ruibo Zhang, Saugato Rahman Dhruba, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Subjects
Science - Abstract
Convolutional Neural Networks (CNN) are often unsuitable for predictive modeling involving nonimage based biological features. Here, the authors present a mapping termed REFINED to represent high dimensional vectors as compact images with spatial correlation that makes it compatible with CNN based learning.
- Published
- 2020
- Full Text
- View/download PDF
4. Probabilistic modeling of personalized drug combinations from integrated chemical screen and molecular data in sarcoma
- Author
-
Noah E. Berlow, Rishi Rikhi, Mathew Geltzeiler, Jinu Abraham, Matthew N. Svalina, Lara E. Davis, Erin Wise, Maria Mancini, Jonathan Noujaim, Atiya Mansoor, Michael J. Quist, Kevin L. Matlock, Martin W. Goros, Brian S. Hernandez, Yee C. Doung, Khin Thway, Tomohide Tsukahara, Jun Nishio, Elaine T. Huang, Susan Airhart, Carol J. Bult, Regina Gandour-Edwards, Robert G. Maki, Robin L. Jones, Joel E. Michalek, Milan Milovancev, Souparno Ghosh, Ranadip Pal, and Charles Keller
- Subjects
Personalized therapy ,Combination therapy ,Artificial intelligence and machine learning ,Pediatric cancer ,Sarcoma ,Drug screening ,Neoplasms. Tumors. Oncology. Including cancer and carcinogens ,RC254-282 - Abstract
Abstract Background Cancer patients with advanced disease routinely exhaust available clinical regimens and lack actionable genomic medicine results, leaving a large patient population without effective treatments options when their disease inevitably progresses. To address the unmet clinical need for evidence-based therapy assignment when standard clinical approaches have failed, we have developed a probabilistic computational modeling approach which integrates molecular sequencing data with functional assay data to develop patient-specific combination cancer treatments. Methods Tissue taken from a murine model of alveolar rhabdomyosarcoma was used to perform single agent drug screening and DNA/RNA sequencing experiments; results integrated via our computational modeling approach identified a synergistic personalized two-drug combination. Cells derived from the primary murine tumor were allografted into mouse models and used to validate the personalized two-drug combination. Computational modeling of single agent drug screening and RNA sequencing of multiple heterogenous sites from a single patient’s epithelioid sarcoma identified a personalized two-drug combination effective across all tumor regions. The heterogeneity-consensus combination was validated in a xenograft model derived from the patient’s primary tumor. Cell cultures derived from human and canine undifferentiated pleomorphic sarcoma were assayed by drug screen; computational modeling identified a resistance-abrogating two-drug combination common to both cell cultures. This combination was validated in vitro via a cell regrowth assay. Results Our computational modeling approach addresses three major challenges in personalized cancer therapy: synergistic drug combination predictions (validated in vitro and in vivo in a genetically engineered murine cancer model), identification of unifying therapeutic targets to overcome intra-tumor heterogeneity (validated in vivo in a human cancer xenograft), and mitigation of cancer cell resistance and rewiring mechanisms (validated in vitro in a human and canine cancer model). Conclusions These proof-of-concept studies support the use of an integrative functional approach to personalized combination therapy prediction for the population of high-risk cancer patients lacking viable clinical options and without actionable DNA sequencing-based therapy.
- Published
- 2019
- Full Text
- View/download PDF
5. Recursive model for dose-time responses in pharmacological studies
- Author
-
Saugato Rahman Dhruba, Aminur Rahman, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Subjects
Drug sensitivity prediction ,Pharmacogenomic studies ,HMS-LINCS ,Joint dose-time modeling ,Recursive modeling ,Dose-response curve ,Computer applications to medicine. Medical informatics ,R858-859.7 ,Biology (General) ,QH301-705.5 - Abstract
Abstract Background Clinical studies often track dose-response curves of subjects over time. One can easily model the dose-response curve at each time point with Hill equation, but such a model fails to capture the temporal evolution of the curves. On the other hand, one can use Gompertz equation to model the temporal behaviors at each dose without capturing the evolution of time curves across dosage. Results In this article, we propose a parametric model for dose-time responses that follows Gompertz law in time and Hill equation across dose approximately. We derive a recursion relation for dose-response curves over time capturing the temporal evolution and then specify a regression model connecting the parameters controlling the dose-time responses with individual level proteomic data. The resultant joint model allows us to predict the dose-response curves over time for new individuals. Conclusion We have compared the efficacy of our proposed Recursive Hybrid model with individual dose-response predictive models at desired time points. We note that our proposed model exhibits a superior performance compared to the individual ones for both synthetic data and actual pharmacological data. For the desired dose-time varying genetic characterization and drug response values, we have used the HMS-LINCS database and demonstrated the effectiveness of our model for all available anticancer compounds.
- Published
- 2019
- Full Text
- View/download PDF
6. Application of transfer learning for cancer drug sensitivity prediction
- Author
-
Saugato Rahman Dhruba, Raziur Rahman, Kevin Matlock, Souparno Ghosh, and Ranadip Pal
- Subjects
Drug sensitivity prediction ,Pharmacogenomic studies ,CCLE ,GDSC ,Transfer learning ,Nonlinear mapping ,Computer applications to medicine. Medical informatics ,R858-859.7 ,Biology (General) ,QH301-705.5 - Abstract
Abstract Background In precision medicine, scarcity of suitable biological data often hinders the design of an appropriate predictive model. In this regard, large scale pharmacogenomics studies, like CCLE and GDSC hold the promise to mitigate the issue. However, one cannot directly employ data from multiple sources together due to the existing distribution shift in data. One way to solve this problem is to utilize the transfer learning methodologies tailored to fit in this specific context. Results In this paper, we present two novel approaches for incorporating information from a secondary database for improving the prediction in a target database. The first approach is based on latent variable cost optimization and the second approach considers polynomial mapping between the two databases. Utilizing CCLE and GDSC databases, we illustrate that the proposed approaches accomplish a better prediction of drug sensitivities for different scenarios as compared to the existing approaches. Conclusion We have compared the performance of the proposed predictive models with database-specific individual models as well as existing transfer learning approaches. We note that our proposed approaches exhibit superior performance compared to the abovementioned alternative techniques for predicting sensitivity for different anti-cancer compounds, particularly the nonlinear mapping model shows the best overall performance.
- Published
- 2018
- Full Text
- View/download PDF
7. Investigation of model stacking for drug sensitivity prediction
- Author
-
Kevin Matlock, Carlos De Niz, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Subjects
Drug sensitivity prediction ,Stacking ,Bias ,Computer applications to medicine. Medical informatics ,R858-859.7 ,Biology (General) ,QH301-705.5 - Abstract
Abstract Background A significant problem in precision medicine is the prediction of drug sensitivity for individual cancer cell lines. Predictive models such as Random Forests have shown promising performance while predicting from individual genomic features such as gene expressions. However, accessibility of various other forms of data types including information on multiple tested drugs necessitates the examination of designing predictive models incorporating the various data types. Results We explore the predictive performance of model stacking and the effect of stacking on the predictive bias and squared error. In addition we discuss the analytical underpinnings supporting the advantages of stacking in reducing squared error and inherent bias of random forests in prediction of outliers. The framework is tested on a setup including gene expression, drug target, physical properties and drug response information for a set of drugs and cell lines. Conclusion The performance of individual and stacked models are compared. We note that stacking models built on two heterogeneous datasets provide superior performance to stacking different models built on the same dataset. It is also noted that stacking provides a noticeable reduction in the bias of our predictors when the dominant eigenvalue of the principle axis of variation in the residuals is significantly higher than the remaining eigenvalues.
- Published
- 2018
- Full Text
- View/download PDF
8. Design of Probabilistic Random Forests with Applications to Anticancer Drug Sensitivity Prediction
- Author
-
Raziur Rahman, Saad Haider, Souparno Ghosh, and Ranadip Pal
- Subjects
Neoplasms. Tumors. Oncology. Including cancer and carcinogens ,RC254-282 - Published
- 2016
9. A Copula Based Approach for Design of Multivariate Random Forests for Drug Sensitivity Prediction.
- Author
-
Saad Haider, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Subjects
Medicine ,Science - Abstract
Modeling sensitivity to drugs based on genetic characterizations is a significant challenge in the area of systems medicine. Ensemble based approaches such as Random Forests have been shown to perform well in both individual sensitivity prediction studies and team science based prediction challenges. However, Random Forests generate a deterministic predictive model for each drug based on the genetic characterization of the cell lines and ignores the relationship between different drug sensitivities during model generation. This application motivates the need for generation of multivariate ensemble learning techniques that can increase prediction accuracy and improve variable importance ranking by incorporating the relationships between different output responses. In this article, we propose a novel cost criterion that captures the dissimilarity in the output response structure between the training data and node samples as the difference in the two empirical copulas. We illustrate that copulas are suitable for capturing the multivariate structure of output responses independent of the marginal distributions and the copula based multivariate random forest framework can provide higher accuracy prediction and improved variable selection. The proposed framework has been validated on genomics of drug sensitivity for cancer and cancer cell line encyclopedia database.
- Published
- 2015
- Full Text
- View/download PDF
10. Design of Probabilistic Random Forests with Applications to Anticancer Drug Sensitivity Prediction
- Author
-
Raziur Rahman, Saad Haider, Souparno Ghosh, and Ranadip Pal
- Subjects
Neoplasms. Tumors. Oncology. Including cancer and carcinogens ,RC254-282 - Abstract
Random forests consisting of an ensemble of regression trees with equal weights are frequently used for design of predictive models. In this article, we consider an extension of the methodology by representing the regression trees in the form of probabilistic trees and analyzing the nature of heteroscedasticity. The probabilistic tree representation allows for analytical computation of confidence intervals (CIs), and the tree weight optimization is expected to provide stricter CIs with comparable performance in mean error. We approached the ensemble of probabilistic trees’ prediction from the perspectives of a mixture distribution and as a weighted sum of correlated random variables. We applied our methodology to the drug sensitivity prediction problem on synthetic and cancer cell line encyclopedia dataset and illustrated that tree weights can be selected to reduce the average length of the CI without increase in mean error.
- Published
- 2015
- Full Text
- View/download PDF
11. Efficient Normalized Conformal Prediction and Uncertainty Quantification for Anti-Cancer Drug Sensitivity Prediction with Deep Regression Forests.
- Author
-
Daniel Nolte, Souparno Ghosh, and Ranadip Pal
- Published
- 2024
- Full Text
- View/download PDF
12. Sparse Bayesian variable selection in high-dimensional logistic regression models with correlated priors.
- Author
-
Zhuanzhuan Ma, Zifei Han, Souparno Ghosh, Liucang Wu, and Min Wang 0004
- Published
- 2024
- Full Text
- View/download PDF
13. A novel Bayesian method for variable selection and estimation in binary quantile regression.
- Author
-
Mai Dao, Min Wang 0004, and Souparno Ghosh
- Published
- 2022
- Full Text
- View/download PDF
14. Bayesian variable selection and estimation in quantile regression using a quantile-specific prior.
- Author
-
Mai Dao, Min Wang 0004, Souparno Ghosh, and Keying Ye
- Published
- 2022
- Full Text
- View/download PDF
15. A Matrix Ensemble Kalman Filter-based Multi-arm Neural Network to Adequately Approximate Deep Neural Networks.
- Author
-
Ved Piyush, Yuchen Yan, Yuzhen Zhou, Yanbin Yin, and Souparno Ghosh
- Published
- 2023
- Full Text
- View/download PDF
16. Investigation of REFINED CNN ensemble learning for anti-cancer drug sensitivity prediction.
- Author
-
Omid Bazgir, Souparno Ghosh, and Ranadip Pal
- Published
- 2021
- Full Text
- View/download PDF
17. A spatially explicit model of postdisaster housing recovery.
- Author
-
Ali Nejat, Roxana J. Javid, Souparno Ghosh, and Saeed Moradi
- Published
- 2020
- Full Text
- View/download PDF
18. Predicting binding affinities of emerging variants of SARS-CoV-2 using spike protein sequencing data: observations, caveats and recommendations.
- Author
-
Ruibo Zhang, Souparno Ghosh, and Ranadip Pal
- Published
- 2022
- Full Text
- View/download PDF
19. Dimensionality Reduction based Transfer Learning applied to Pharmacogenomics Databases.
- Author
-
Saugato Rahman Dhruba, Raziur Rahman, Kevin Matlock, Souparno Ghosh, and Ranadip Pal
- Published
- 2018
- Full Text
- View/download PDF
20. Adaptive Multi-task Elastic Net based feature selection from Pharmacogenomics Databases.
- Author
-
Raziur Rahman, Chamila Perera, Souparno Ghosh, and Ranadip Pal
- Published
- 2018
- Full Text
- View/download PDF
21. Social media data and post-disaster recovery.
- Author
-
Mehdi Jamali, Ali Nejat, Souparno Ghosh, Fang Jin, and Guofeng Cao
- Published
- 2019
- Full Text
- View/download PDF
22. Evaluating the consistency of large-scale pharmacogenomic studies.
- Author
-
Raziur Rahman, Saugato Rahman Dhruba, Kevin Matlock, Carlos De Niz, Souparno Ghosh, and Ranadip Pal
- Published
- 2019
- Full Text
- View/download PDF
23. Investigation of REFINED CNN ensemble learning for anti-cancer drug sensitivity prediction.
- Author
-
Omid Bazgir, Souparno Ghosh, and Ranadip Pal
- Published
- 2020
24. Sequential feature selection and inference using multi-variate random forests.
- Author
-
Joshua Mayer, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Published
- 2018
- Full Text
- View/download PDF
25. Sstack: an R package for stacking with applications to scenarios involving sequential addition of samples and features.
- Author
-
Kevin Matlock, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Published
- 2019
- Full Text
- View/download PDF
26. Probing nitric oxide signaling using molecular MRI
- Author
-
Ali Barandov, Souparno Ghosh, and Alan Jasanoff
- Subjects
Molecular Probes ,Physiology (medical) ,Animals ,Contrast Media ,Humans ,Nitric Oxide Synthase ,Nitric Oxide ,Magnetic Resonance Imaging ,Biochemistry - Abstract
Wide field measurements of nitric oxide (NO) signaling could help understand and diagnose the many physiological processes in which NO plays a key role. Magnetic resonance imaging (MRI) can support particularly powerful approaches for this purpose if equipped with molecular probes sensitized to NO and NO-associated targets. In this review, we discuss the development of MRI-detectable probes that could enable studies of nitrergic signaling in animals and potentially human subjects. Major families of probes include contrast agents designed to capture and report integrated NO levels directly, as well as molecules that respond to or emulate the activity of nitric oxide synthase enzymes. For each group, we outline the relevant molecular mechanisms and discuss results that have been obtained in vitro and in animals. The most promising in vivo data described to date have been acquired using NO capture-based relaxation agents and using engineered nitric oxide synthases that provide hemodynamic readouts of NO signaling pathway activation. These advances establish a beachhead for ongoing efforts to improve the sensitivity, specificity, and clinical applicability of NO-related molecular MRI technology.
- Published
- 2022
27. REFINED (REpresentation of Features as Images with NEighborhood Dependencies): A novel feature representation for Convolutional Neural Networks.
- Author
-
Omid Bazgir, Ruibo Zhang, Saugato Rahman Dhruba, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Published
- 2019
28. Federated learning framework integrating REFINED CNN and Deep Regression Forests
- Author
-
Daniel Nolte, Omid Bazgir, Souparno Ghosh, and Ranadip Pal
- Subjects
General Medicine - Abstract
Summary Predictive learning from medical data incurs additional challenge due to concerns over privacy and security of personal data. Federated learning, intentionally structured to preserve high level of privacy, is emerging to be an attractive way to generate cross-silo predictions in medical scenarios. However, the impact of severe population-level heterogeneity on federated learners is not well explored. In this article, we propose a methodology to detect presence of population heterogeneity in federated settings and propose a solution to handle such heterogeneity by developing a federated version of Deep Regression Forests. Additionally, we demonstrate that the recently conceptualized REpresentation of Features as Images with NEighborhood Dependencies CNN framework can be combined with the proposed Federated Deep Regression Forests to provide improved performance as compared to existing approaches. Availability and implementation The Python source code for reproducing the main results are available on GitHub: https://github.com/DanielNolte/FederatedDeepRegressionForests. Contact ranadip.pal@ttu.edu Supplementary information Supplementary data are available at Bioinformatics Advances online.
- Published
- 2023
29. Recursive Model for Dose-time Responses in Pharmacological Studies.
- Author
-
Aminur Rahman, Saugato Rahman Dhruba, Souparno Ghosh, and Ranadip Pal
- Published
- 2018
- Full Text
- View/download PDF
30. A multivariate regression approach toward prioritizing BIM adoption barriers in the Ethiopian construction industry
- Author
-
Shihunegn Alemayehu, Ali Nejat, Souparno Ghosh, and Tewodros Ghebrab
- Subjects
Construction management ,Multivariate statistics ,Process management ,020209 energy ,0211 other engineering and technologies ,02 engineering and technology ,Building and Construction ,General Business, Management and Accounting ,Construction industry ,021105 building & construction ,Architecture ,0202 electrical engineering, electronic engineering, information engineering ,Business ,Civil and Structural Engineering - Abstract
PurposeBuilding information modeling (BIM) is a process of creating an intelligent virtual model integrating project data from design to construction and operation. BIM models enhance the process of communicating the progress of construction to stakeholders and facilitate integrated project delivery, coordination and clash detection. However, barriers within the construction industry in Ethiopia has led to slow BIM adoption in the country. The aim of this paper is to identify perceived BIM barriers, provide a platform to quantify their importance and develop a regression model to link individual's personal/professional attributes to their perception of BIM barrier.Design/methodology/approachTo address the objectives of this research, an online survey was developed to collect feedback from construction professionals in Ethiopia on 20 major adoption barriers extracted from a thorough review of literature. Relative importance index and strength of consensus metric were employed to identify the significance of barriers. This was then succeeded by performing exploratory factor analysis to determine the major constructs of BIM barriers which was then used to develop a multivariate regression model linking respondents' personal attributes to their perception of BIM barrier.FindingsResults revealed the importance of project complexity and BIM maturity level in prioritizing barriers that are more relevant under various contexts. More specifically, results indicated the following study highlights: Project complexity led to higher perceived weights for lack of appropriate physical/cloud infrastructures, and a BIM standard. Higher levels of BIM maturity signified the importance of BIM internal issues such as liability, licensing and maintenance issues among other adoption barriers. Female participants tended not to consider intangibility of BIM benefits as a major barrier towards BIM adoption compared to male participants. Age of the participants turned out to be the least important factor in their prioritization of BIM perceived adoption barriers.Originality/valueWhile many research studies have explored BIM adoption barriers in various countries around the world, none to the best of the authors' knowledge have attempted to develop a model to highlight the impact of individuals' personal/professional attributes on their perception of adoption barriers within their community which can help with prioritizing the barriers that are deemed to be more important given the characteristics of the community under study. Our result indicated the importance of BIM maturity level and project complexity in prioritizing barriers associated with BIM adoption within Ethiopia's construction industry.
- Published
- 2021
31. Active Shooter Detection in Multiple-Person Scenario Using RF-Based Machine Vision
- Author
-
Ranadip Pal, Changzhi Li, Daniel Nolte, Souparno Ghosh, Yiran Li, Saugato Rahman Dhruba, and Omid Bazgir
- Subjects
business.industry ,Computer science ,Machine vision ,Deep learning ,010401 analytical chemistry ,Bayesian optimization ,Feature extraction ,01 natural sciences ,Convolutional neural network ,0104 chemical sciences ,Gesture recognition ,Spectrogram ,Anomaly detection ,Computer vision ,Artificial intelligence ,Electrical and Electronic Engineering ,business ,Instrumentation - Abstract
Emerging applications of radio frequency (RF) vision sensors for security and gesture recognition primarily target single individual scenarios which restricts potential applications. In this article, we present the design of a cyber-physical framework that analyzes RF micro-Doppler signatures for individual anomaly detection, such as a hidden rifle among multiple individuals. RF avoids certain limitations of video surveillance, such as recognizing concealed objects and privacy concerns. Current RF-based approaches for human activity detection or gesture recognition usually consider single individual scenarios, and the features extracted for such scenarios are not applicable for multi-person cases. From a machine learning perspective, the RF sensor spectrogram images are conducible for training using deep convolutional neural networks. However, generating a large labeled training dataset with an exhaustive variety of multi-person scenarios is extremely time consuming and nearly impossible due to the wide range of combinations possible. We present approaches for multi-person spectrogram generation based on individual person spectrograms that can augment the training dataset and increase the accuracy of prediction. Our results show that the spectrogram generated by RF sensors can be harnessed by artificial intelligence algorithms to detect anomalies such as a concealed weapon for single and multiple people scenarios. The proposed system can aid as a standalone tool, or be complemented by video surveillance for anomaly detection, in scenarios involving single or multiple individuals.
- Published
- 2021
32. Investigation of Model Stacking for Drug Sensitivity Prediction.
- Author
-
Kevin Matlock, Carlos De Niz, Raziur Rahman, Souparno Ghosh, and Ranadip Pal
- Published
- 2017
- Full Text
- View/download PDF
33. Predicting binding affinities of emerging variants of SARS-CoV-2 using spike protein sequencing data: observations, caveats and recommendations
- Author
-
Ruibo, Zhang, Souparno, Ghosh, and Ranadip, Pal
- Subjects
Proteomics ,SARS-CoV-2 ,Sequence Analysis, Protein ,Spike Glycoprotein, Coronavirus ,COVID-19 ,Humans ,Amino Acid Sequence ,Molecular Biology ,Protein Binding ,Information Systems - Abstract
Predicting protein properties from amino acid sequences is an important problem in biology and pharmacology. Protein–protein interactions among SARS-CoV-2 spike protein, human receptors and antibodies are key determinants of the potency of this virus and its ability to evade the human immune response. As a rapidly evolving virus, SARS-CoV-2 has already developed into many variants with considerable variation in virulence among these variants. Utilizing the proteomic data of SARS-CoV-2 to predict its viral characteristics will, therefore, greatly aid in disease control and prevention. In this paper, we review and compare recent successful prediction methods based on long short-term memory (LSTM), transformer, convolutional neural network (CNN) and a similarity-based topological regression (TR) model and offer recommendations about appropriate predictive methodology depending on the similarity between training and test datasets. We compare the effectiveness of these models in predicting the binding affinity and expression of SARS-CoV-2 spike protein sequences. We also explore how effective these predictive methods are when trained on laboratory-created data and are tasked with predicting the binding affinity of the in-the-wild SARS-CoV-2 spike protein sequences obtained from the GISAID datasets. We observe that TR is a better method when the sample size is small and test protein sequences are sufficiently similar to the training sequence. However, when the training sample size is sufficiently large and prediction requires extrapolation, LSTM embedding and CNN-based predictive model show superior performance.
- Published
- 2022
34. Functional dissection of neural circuitry using a genetic reporter for fMRI
- Author
-
Souparno Ghosh, Nan Li, Miriam Schwalm, Benjamin B. Bartelle, Tianshu Xie, Jade I. Daher, Urvashi D. Singh, Katherine Xie, Nicholas DiNapoli, Nicholas B. Evans, Kwanghun Chung, and Alan Jasanoff
- Subjects
Mammals ,Brain Mapping ,Reward ,General Neuroscience ,Animals ,Brain ,Magnetic Resonance Imaging ,Article ,Corpus Striatum ,Rats - Abstract
The complex connectivity of the mammalian brain underlies its function, but understanding how interconnected brain regions interact in neural processing remains a formidable challenge. Here we address this problem by introducing a genetic probe that permits selective functional imaging of distributed neural populations defined by viral labeling techniques. The probe is an engineered enzyme that transduces cytosolic calcium dynamics of probe-expressing cells into localized hemodynamic responses that can be specifically visualized by functional magnetic resonance imaging. Using a viral vector that undergoes retrograde transport, we apply the probe to characterize a brain-wide network of presynaptic inputs to the striatum activated in a deep brain stimulation paradigm in rats. The results reveal engagement of surprisingly diverse projection sources and inform an integrated model of striatal function relevant to reward behavior and therapeutic neurostimulation approaches. Our work thus establishes a strategy for mechanistic analysis of multiregional neural systems in the mammalian brain.
- Published
- 2022
35. Deep Learning Methods for Tassel Count Time-Series
- Author
-
Gayara Fernando, Ved Piyush, and Souparno Ghosh
- Published
- 2022
36. Mitochondrial dysfunction triggers secretion of the immunosuppressive factor α-fetoprotein
- Author
-
Mochoruk K, Zakery N. Baker, Savard C, Aren Boulet, DeCoteau J, Christopher Lowden, Brendan J. Battersby, Scot C. Leary, Harry Cheng, Yuan S, Kimberly A. Jett, Ahmed Hossain, Ioannou G, Paul A. Cobine, Yilmaz O, Souparno Ghosh, Ng P, Brian N. Kim, Gohil Vmm, and Barretto K
- Subjects
0303 health sciences ,Cell type ,Leukopenia ,Barth syndrome ,Biology ,medicine.disease ,3. Good health ,03 medical and health sciences ,0302 clinical medicine ,Immune system ,medicine.anatomical_structure ,Downregulation and upregulation ,Cell surface receptor ,White blood cell ,medicine ,Cancer research ,Secretion ,medicine.symptom ,030217 neurology & neurosurgery ,030304 developmental biology - Abstract
Signaling circuits crucial to systemic physiology are widespread, yet uncovering their molecular underpinnings remains a barrier to understanding the etiology of many metabolic disorders. Here, we identify a copper-linked signaling circuit activated by disruption of mitochondrial function in the murine liver or heart that results in atrophy of the spleen and thymus and causes a peripheral white blood cell deficiency. We demonstrate that the leukopenia is caused by α-fetoprotein, which requires copper and the cell surface receptor CCR5 to promote white blood cell death. We further show that α-fetoprotein expression is upregulated in several cell types upon inhibition of oxidative phosphorylation, including a muscle cell model of Barth syndrome. Collectively, our data argue that α-fetoprotein secreted by bioenergetically stressed tissue suppresses the immune system, an effect which may explain the recurrent infections that are observed in a subset of mitochondrial diseases or in other disorders with mitochondrial involvement.
- Published
- 2021
37. Social media data and post-disaster recovery
- Author
-
Fang Jin, Souparno Ghosh, Ali Nejat, Guofeng Cao, and Mehdi Jamali
- Subjects
Disaster experience ,Computer Networks and Communications ,05 social sciences ,Context (language use) ,02 engineering and technology ,Library and Information Sciences ,Data science ,Identification (information) ,020204 information systems ,0502 economics and business ,0202 electrical engineering, electronic engineering, information engineering ,050211 marketing ,Social media ,Psychology ,Post disaster ,Information Systems ,Recovery phase - Abstract
This study introduces a multi-step methodology for analyzing social media data during the post-disaster recovery phase of Hurricane Sandy. Its outputs include identification of the people who experienced the disaster, estimates of their physical location, assessments of the topics they discussed post-disaster, analysis of the tract-level relationships between the topics people discussed and tract-level internal attributes, and a comparison of these outputs to those of people who did not experience the disaster. Faith-based, community, assets, and financial topics emerged as major topics of discussion within the context of the disaster experience. The differences between predictors of these topics compared to those of people who did not experience the disaster were investigated in depth, revealing considerable differences among vulnerable populations. The use of this methodology as a new Machine Learning Algorithm to analyze large volumes of social media data is advocated in the conclusion.
- Published
- 2019
38. Functional dissection of neural circuitry using a genetic reporter for fMRI
- Author
-
Souparno Ghosh, Kwanghun Chung, Alan Jasanoff, Nicholas B. Evans, Katherine Xie, Benjamin B. Bartelle, Nan Li, Nicholas DiNapoli, Urvashi D. Singh, Jade I. Daher, and Tianshu Xie
- Subjects
Functional imaging ,Deep brain stimulation ,medicine.diagnostic_test ,Computer science ,medicine.medical_treatment ,medicine ,Axoplasmic transport ,Biological neural network ,Striatum ,Functional magnetic resonance imaging ,Neuroscience ,Neurostimulation ,Function (biology) - Abstract
The complex connectivity of the mammalian brain underlies its function, but understanding how interconnected brain regions interact in neural processing remains a formidable challenge. Here we address this problem by introducing a genetic probe that permits selective functional imaging of neural circuit elements defined by their synaptic interrelationships throughout the brain. The probe is an engineered enzyme that transduces cytosolic calcium dynamics of probe-expressing cells into localized hemodynamic responses that can be selectively visualized by functional magnetic resonance imaging. Using a viral vector that undergoes retrograde transport, we apply the probe to characterize a brain-wide network of monosynaptic inputs to the striatum activated in a deep brain stimulation paradigm in rats. The results reveal engagement of surprisingly diverse projection sources and inform an integrated model of striatal function relevant to reward behavior and therapeutic neurostimulation approaches. Our work thus establishes a potent strategy for mechanistic analysis of distributed neural systems.
- Published
- 2020
39. Representation of features as images with neighborhood dependencies for compatibility with convolutional neural networks
- Author
-
Ruibo Zhang, Raziur Rahman, Omid Bazgir, Saugato Rahman Dhruba, Souparno Ghosh, and Ranadip Pal
- Subjects
0301 basic medicine ,Computer science ,Science ,Computer Science::Neural and Evolutionary Computation ,Feature extraction ,Bayesian probability ,Datasets as Topic ,General Physics and Astronomy ,Antineoplastic Agents ,Image processing ,02 engineering and technology ,Convolutional neural network ,Article ,General Biochemistry, Genetics and Molecular Biology ,03 medical and health sciences ,Bayes' theorem ,Deep Learning ,Gene expression analysis ,Targeted therapies ,Cell Line, Tumor ,Neoplasms ,Machine learning ,Biomarkers, Tumor ,Image Processing, Computer-Assisted ,Humans ,Multidimensional scaling ,lcsh:Science ,Cell Proliferation ,Oligonucleotide Array Sequence Analysis ,Multidisciplinary ,business.industry ,Gene Expression Profiling ,Deep learning ,High-Throughput Nucleotide Sequencing ,Bayes Theorem ,Pattern recognition ,General Chemistry ,021001 nanoscience & nanotechnology ,030104 developmental biology ,Drug Resistance, Neoplasm ,Computer Science::Computer Vision and Pattern Recognition ,lcsh:Q ,Pairwise comparison ,Artificial intelligence ,Drug Screening Assays, Antitumor ,0210 nano-technology ,business - Abstract
Deep learning with Convolutional Neural Networks has shown great promise in image-based classification and enhancement but is often unsuitable for predictive modeling using features without spatial correlations. We present a feature representation approach termed REFINED (REpresentation of Features as Images with NEighborhood Dependencies) to arrange high-dimensional vectors in a compact image form conducible for CNN-based deep learning. We consider the similarities between features to generate a concise feature map in the form of a two-dimensional image by minimizing the pairwise distance values following a Bayesian Metric Multidimensional Scaling Approach. We hypothesize that this approach enables embedded feature extraction and, integrated with CNN-based deep learning, can boost the predictive accuracy. We illustrate the superior predictive capabilities of the proposed framework as compared to state-of-the-art methodologies in drug sensitivity prediction scenarios using synthetic datasets, drug chemical descriptors as predictors from NCI60, and both transcriptomic information and drug descriptors as predictors from GDSC., Convolutional Neural Networks (CNN) are often unsuitable for predictive modeling involving nonimage based biological features. Here, the authors present a mapping termed REFINED to represent high dimensional vectors as compact images with spatial correlation that makes it compatible with CNN based learning.
- Published
- 2020
40. Tumor ablation due to inhomogeneous anisotropic diffusion in generic three-dimensional topologies
- Author
-
Souparno Ghosh, Erdi Kara, Aminur Rahman, and Eugenio Aulisa
- Subjects
Physics ,Tumor region ,Anisotropic diffusion ,Diffusion tensor magnetic resonance imaging ,Network topology ,01 natural sciences ,Tumor ablation ,Finite element method ,010305 fluids & plasmas ,Diffusion Tensor Imaging ,Neoplasms ,0103 physical sciences ,Full model ,Anisotropy ,Humans ,Statistical physics ,Diffusion (business) ,010306 general physics - Abstract
In recent decades computer-aided technologies have become prevalent in medicine, however, cancer drugs are often only tested on in vitro cell lines from biopsies. We derive a full three-dimensional model of inhomogeneous -anisotropic diffusion in a tumor region coupled to a binary population model, which simulates in vivo scenarios faster than traditional cell-line tests. The diffusion tensors are acquired using diffusion tensor magnetic resonance imaging from a patient diagnosed with glioblastoma multiform. Then we numerically simulate the full model with finite element methods and produce drug concentration heat maps, apoptosis hotspots, and dose-response curves. Finally, predictions are made about optimal injection locations and volumes, which are presented in a form that can be employed by doctors and oncologists.
- Published
- 2020
41. Simulation of neutron background for a dark matter search experiment at JUSL
- Author
-
Samrat Dutta, Souparno Ghosh, S. Banik, Bedangadas Mohanty, Satyajit Saha, P. Bhattacharjee, V.K.S. Kashyap, and K. K. Meghna
- Subjects
Physics ,Muon ,010308 nuclear & particles physics ,Astrophysics::High Energy Astrophysical Phenomena ,Dark matter ,FOS: Physical sciences ,Radiation ,01 natural sciences ,030218 nuclear medicine & medical imaging ,Physics::Geophysics ,Nuclear physics ,03 medical and health sciences ,0302 clinical medicine ,WIMP ,0103 physical sciences ,Neutron source ,Neutron ,Astrophysics - Instrumentation and Methods for Astrophysics ,Instrumentation ,Axion ,Instrumentation and Methods for Astrophysics (astro-ph.IM) ,Mathematical Physics ,Spontaneous fission - Abstract
Dark matter search experiments demand a low to ultralow radiation background to operate. It is very important to understand the nature of the radiation background including knowledge about the sources contributing to it. Sometimes, evaluation of the background becomes very specific to the site chosen for the experiment, and also to the experimental configuration. A dark matter search experiment is proposed to be set up at the Jaduguda Underground Science Laboratory (JUSL) in India. The laboratory will be located inside an existing mine with 555 m of vertical rock overburden. Neutrons produced from (α,n) reactions, spontaneous fission of natural radioactive impurities in the rocks, and also from cosmic muon-induced reactions are considered as the main background which can affect the sensitivity and outcome of the experiment. In this work, simulations based on GEANT4 are done to understand both the radiogenic neutron background caused by natural radioactivity of the surrounding rock and the cosmogenic neutron background due to interactions of the deeply penetrating cosmic muons with the rock material. The muon flux in the cavern is obtained to be 4.49(±0.25)×10-7 cm-2s-1 and the fluxes of radiogenic and cosmogenic neutrons above an energy threshold of 1 MeV in the cavern are obtained to be 5.75(±0.69)×10-6cm-2s-1 and 7.25(±0.40)× 10-9cm-2 s-1 respectively. The values obtained are comparable with estimates and measurements from DAMA, WIPP, and dark matter experiments at Boulby mine. The effectiveness of different shielding materials is also investigated to obtain the best possible neutron background reduction for a dark matter search experiment at JUSL. We also estimate the sensitivity of a CsI based detector for Weakly Interacting Massive Particle (WIMP) dark matter search at JUSL considering the estimated neutron background.
- Published
- 2020
- Full Text
- View/download PDF
42. Molecular Magnetic Resonance Imaging of Nitric Oxide in Biological Systems
- Author
-
Benjamin B. Bartelle, Jade I. Daher, Hannah Collins, Ali Barandov, Michael L. Pegis, Alan Jasanoff, Souparno Ghosh, and Nan Li
- Subjects
Contrast Media ,Bioengineering ,Endogeny ,02 engineering and technology ,Nitric Oxide ,01 natural sciences ,Article ,Nitric oxide ,chemistry.chemical_compound ,In vivo ,medicine ,Animals ,Instrumentation ,Reactive nitrogen species ,Neuroinflammation ,Fluid Flow and Transfer Processes ,medicine.diagnostic_test ,biology ,Chemistry ,Process Chemistry and Technology ,010401 analytical chemistry ,Magnetic resonance imaging ,021001 nanoscience & nanotechnology ,Magnetic Resonance Imaging ,In vitro ,0104 chemical sciences ,Rats ,Nitric oxide synthase ,Oxygen ,Biophysics ,biology.protein ,Nitric Oxide Synthase ,0210 nano-technology - Abstract
Copyright © 2020 American Chemical Society. Detection of nitric oxide (NO) in biological systems is challenging due to both physicochemical properties of NO and limitations of current imaging modalities and probes. Magnetic resonance imaging (MRI) could be applied for studying NO in living tissue with high spatiotemporal resolution, but there is still a need for chemical agents that effectively sensitize MRI to biological NO production. To develop a suitable probe, we studied the interactions between NO and a library of manganese complexes with various oxidation states and molecular structures. Among this set, the manganese(III) complex with N,N′-(1,2-phenylene)bis(5-fluoro-2-hydroxybenzamide) showed favorable changes in longitudinal relaxivity upon addition of NO-releasing chemicals in vitro while also maintaining selectivity against other biologically relevant reactive nitrogen and oxygen species, making it a suitable NO-responsive contrast agent for T1-weighted MRI. When loaded with this compound, cells ectopically expressing nitric oxide synthase (NOS) isoforms showed MRI signal decreases of over 20% compared to control cells and were also responsive to NOS inhibition or calcium-dependent activation. The sensor could also detect endogenous NOS activity in antigen-stimulated macrophages and in a rat model of neuroinflammation in vivo. Given the key role of NO and associated reactive nitrogen species in numerous physiological and pathological processes, MRI approaches based on the new probe could be broadly beneficial for studies of NO-related signaling in living subjects.
- Published
- 2020
43. Anchors of Social Network Awareness Index: A Key to Modeling Postdisaster Housing Recovery
- Author
-
Souparno Ghosh, Ali Nejat, and Saeed Moradi
- Subjects
Index (economics) ,Social network ,business.industry ,Multilevel model ,0211 other engineering and technologies ,020101 civil engineering ,02 engineering and technology ,Environmental economics ,0201 civil engineering ,Domino effect ,Component (UML) ,021105 building & construction ,Key (cryptography) ,business ,Civil and Structural Engineering - Abstract
Reestablishment of housing is a crucial component of the recovery process and has a domino effect on the overall timing of recovery. Anchors of social networks, such as schools and churches...
- Published
- 2019
44. Recursive model for dose-time responses in pharmacological studies
- Author
-
Aminur Rahman, Ranadip Pal, Souparno Ghosh, Raziur Rahman, and Saugato Rahman Dhruba
- Subjects
Time Factors ,Recursive modeling ,Pharmacogenomic studies ,Gompertz function ,lcsh:Computer applications to medicine. Medical informatics ,Biochemistry ,Synthetic data ,03 medical and health sciences ,symbols.namesake ,Drug sensitivity prediction ,0302 clinical medicine ,Structural Biology ,Dose-response curve ,Applied mathematics ,Humans ,Computer Simulation ,Time point ,Molecular Biology ,lcsh:QH301-705.5 ,030304 developmental biology ,Mathematics ,Pharmacology ,0303 health sciences ,Hill differential equation ,Recursion ,Dose-Response Relationship, Drug ,Gompertz law ,Applied Mathematics ,Research ,Regression analysis ,Models, Theoretical ,3. Good health ,Computer Science Applications ,lcsh:Biology (General) ,Databases as Topic ,HMS-LINCS ,030220 oncology & carcinogenesis ,Parametric model ,symbols ,lcsh:R858-859.7 ,Joint dose-time modeling ,Tumor growth model ,Gompertz–Makeham law of mortality - Abstract
Background Clinical studies often track dose-response curves of subjects over time. One can easily model the dose-response curve at each time point with Hill equation, but such a model fails to capture the temporal evolution of the curves. On the other hand, one can use Gompertz equation to model the temporal behaviors at each dose without capturing the evolution of time curves across dosage. Results In this article, we propose a parametric model for dose-time responses that follows Gompertz law in time and Hill equation across dose approximately. We derive a recursion relation for dose-response curves over time capturing the temporal evolution and then specify a regression model connecting the parameters controlling the dose-time responses with individual level proteomic data. The resultant joint model allows us to predict the dose-response curves over time for new individuals. Conclusion We have compared the efficacy of our proposed Recursive Hybrid model with individual dose-response predictive models at desired time points. We note that our proposed model exhibits a superior performance compared to the individual ones for both synthetic data and actual pharmacological data. For the desired dose-time varying genetic characterization and drug response values, we have used the HMS-LINCS database and demonstrated the effectiveness of our model for all available anticancer compounds. Electronic supplementary material The online version of this article (10.1186/s12859-019-2831-4) contains supplementary material, which is available to authorized users.
- Published
- 2019
45. Robust Predictive Model Using Copulas
- Author
-
Quan Hoang, Priyanka Khandelwal, and Souparno Ghosh
- Subjects
Predictive validity ,Computer science ,Copula (linguistics) ,Rejection sampling ,Decision tree ,Econometrics ,Conditional probability distribution ,Regression ,Parametric statistics ,Quantile regression - Abstract
This article presents a robust predictive model using parametric copula-based regression. We show that copula selection test procedures and predictive conditional distributions can be used to assess model adequacy and predictive validity. We offer simulation experiments to demonstrate the ability of our diagnostic procedure to correctly identify the true data generating process. Finally, we apply our methodology on a well-known insurance claims dataset to produce the distribution profile of allocated loss adjustment expense for given pre-specified indemnity payments information. The availability of this entire expense distribution will provide greater insight to the decision-makers before allocating resources for a given insurance claim.
- Published
- 2019
46. Functional random forest with applications in dose-response predictions
- Author
-
Saugato Rahman Dhruba, Ranadip Pal, Raziur Rahman, and Souparno Ghosh
- Subjects
0301 basic medicine ,Multivariate statistics ,Multivariate analysis ,Databases, Pharmaceutical ,Computer science ,lcsh:Medicine ,computer.software_genre ,Article ,Cell Line ,03 medical and health sciences ,0302 clinical medicine ,Neoplasms ,Humans ,Sensitivity (control systems) ,lcsh:Science ,Multidisciplinary ,Dose-Response Relationship, Drug ,lcsh:R ,Univariate ,Reproducibility of Results ,Contrast (statistics) ,Regression analysis ,Models, Theoretical ,3. Good health ,Random forest ,030104 developmental biology ,Area Under Curve ,Multivariate Analysis ,Metric (mathematics) ,Regression Analysis ,lcsh:Q ,Functional regression ,Data mining ,computer ,030217 neurology & neurosurgery - Abstract
Drug sensitivity prediction for individual tumors is a significant challenge in personalized medicine. Current modeling approaches consider prediction of a single metric of the drug response curve such as AUC or IC50. However, the single summary metric of a dose-response curve fails to provide the entire drug sensitivity profile which can be used to design the optimal dose for a patient. In this article, we assess the problem of predicting the complete dose-response curve based on genetic characterizations. We propose an enhancement to the popular ensemble-based Random Forests approach that can directly predict the entire functional profile of a dose-response curve rather than a single summary metric. We design functional regression trees with node costs modified based on dose/response region dependence methodologies and response distribution based approaches. Our results relative to large pharmacological databases such as CCLE and GDSC show a higher accuracy in predicting dose-response curves of the proposed functional framework in contrast to univariate or multivariate Random Forest predicting sensitivities at different dose levels. Furthermore, we also considered the problem of predicting functional responses from functional predictors i.e., estimating the dose-response curves with a model built on dose-dependent expression data. The superior performance of Functional Random Forest using functional data as compared to existing approaches have been shown using the HMS-LINCS dataset. In summary, Functional Random Forest presents an enhanced predictive modeling framework to predict the entire functional response profile considering both static and functional predictors instead of predicting the summary metrics of the response curves.
- Published
- 2019
47. Modeling of drug diffusion in a solid tumor leading to tumor cell death
- Author
-
Aminur Rahman, Ranadip Pal, and Souparno Ghosh
- Subjects
0301 basic medicine ,Quantitative Biology - Subcellular Processes ,Materials science ,Observer (quantum physics) ,Sigmoid function ,Tumor ablation ,3. Good health ,03 medical and health sciences ,030104 developmental biology ,0302 clinical medicine ,FOS: Biological sciences ,030220 oncology & carcinogenesis ,Cell Behavior (q-bio.CB) ,Tumor cell death ,Quantitative Biology - Cell Behavior ,Diffusion (business) ,Solid tumor ,Biological system ,Subcellular Processes (q-bio.SC) ,Free parameter - Abstract
It has been shown recently that changing the fluidic properties of a drug can improve its efficacy in ablating solid tumors. We develop a modeling framework for tumor ablation, and present the simplest possible model for drug diffusion in a spherical tumor with leaky boundaries and assuming cell death eventually leads to ablation of that cell effectively making the two quantities numerically equivalent. The death of a cell after a given exposure time depends on both the concentration of the drug and the amount of oxygen available to the cell. Higher oxygen availability leads to cell death at lower drug concentrations. It can be assumed that a minimum concentration is required for a cell to die, effectively connecting diffusion with efficacy. The concentration threshold decreases as exposure time increases, which allows us to compute dose-response curves. Furthermore, these curves can be plotted at much finer time intervals compared to that of experiments, which is used to produce a dose-threshold-response surface giving an observer a complete picture of the drug's efficacy for an individual. In addition, since the diffusion, leak coefficients, and the availability of oxygen is different for different individuals and tumors, we produce artificial replication data through bootstrapping to simulate error. While the usual data-driven model with Sigmoidal curves use 12 free parameters, our mechanistic model only has two free parameters, allowing it to be open to scrutiny rather than forcing agreement with data. Even so, the simplest model in our framework, derived here, shows close agreement with the bootstrapped curves, and reproduces well established relations, such as Haber's rule., Comment: 25 pages, 65 figures, 2 tables
- Published
- 2018
48. Application of transfer learning for cancer drug sensitivity prediction
- Author
-
Souparno Ghosh, Saugato Rahman Dhruba, Ranadip Pal, Kevin Matlock, and Raziur Rahman
- Subjects
0301 basic medicine ,Polynomial ,Databases, Factual ,Computer science ,Pharmacogenomic studies ,Cancer drugs ,computer.software_genre ,01 natural sciences ,Biochemistry ,Drug sensitivity prediction ,010104 statistics & probability ,Structural Biology ,Neoplasms ,Latent variable ,lcsh:QH301-705.5 ,Biological data ,Applied Mathematics ,3. Good health ,Computer Science Applications ,Gene Expression Regulation, Neoplastic ,Area Under Curve ,lcsh:R858-859.7 ,DNA microarray ,Transfer of learning ,Algorithms ,GDSC ,Antineoplastic Agents ,Context (language use) ,lcsh:Computer applications to medicine. Medical informatics ,Machine learning ,03 medical and health sciences ,CCLE ,Humans ,Cost optimization ,Sensitivity (control systems) ,0101 mathematics ,Molecular Biology ,Nonlinear mapping ,business.industry ,Research ,Scale (chemistry) ,Precision medicine ,Transfer learning ,030104 developmental biology ,lcsh:Biology (General) ,Pharmacogenomics ,Artificial intelligence ,business ,computer - Abstract
Background In precision medicine, scarcity of suitable biological data often hinders the design of an appropriate predictive model. In this regard, large scale pharmacogenomics studies, like CCLE and GDSC hold the promise to mitigate the issue. However, one cannot directly employ data from multiple sources together due to the existing distribution shift in data. One way to solve this problem is to utilize the transfer learning methodologies tailored to fit in this specific context. Results In this paper, we present two novel approaches for incorporating information from a secondary database for improving the prediction in a target database. The first approach is based on latent variable cost optimization and the second approach considers polynomial mapping between the two databases. Utilizing CCLE and GDSC databases, we illustrate that the proposed approaches accomplish a better prediction of drug sensitivities for different scenarios as compared to the existing approaches. Conclusion We have compared the performance of the proposed predictive models with database-specific individual models as well as existing transfer learning approaches. We note that our proposed approaches exhibit superior performance compared to the abovementioned alternative techniques for predicting sensitivity for different anti-cancer compounds, particularly the nonlinear mapping model shows the best overall performance. Electronic supplementary material The online version of this article (10.1186/s12859-018-2465-y) contains supplementary material, which is available to authorized users.
- Published
- 2018
49. Apolipoprotein E interacts with amyloid-β oligomers via positively cooperative multivalent binding
- Author
-
Timir Baran Sil, Souparno Ghosh, Kanchan Garai, and Subhrajyoti Dolai
- Subjects
Gene isoform ,Apolipoprotein E ,chemistry.chemical_compound ,Förster resonance energy transfer ,chemistry ,Amyloid β ,Kinetics ,EDANS ,Biophysics ,Molecule ,lipids (amino acids, peptides, and proteins) ,Multivalent binding - Abstract
Interaction of apolipoprotein E (apoE) isoforms with amyloid-β (Aβ) peptides is considered a critical determinant of the progression of Alzheimer’s disease. However, molecular mechanism of the apoE-Aβ interaction is poorly understood. Here we characterize the nature of the apoE-Aβ complexes and identify the region of apoE that interacts with Aβ. We have prepared three distinct fragments of apoE4,viz., the N-terminal fragment (NTF), hinge domain fragment (HDF) and C-terminal fragment (CTF) to compare its interactions with Aβ. Kinetics of aggregation of Aβ is delayed dramatically in presence of low, substoichiometric concentrations of both NTF and CTF in lipid-free, as well as, in lipidated forms. Effect of HDF is found to be small. Strong inhibition by NTF and CTF at substoichiometric concentrations indicate interactions with the ‘intermediates’ or the oligomers of Aβ. Kinetics of Forster Resonance Energy Transfer (FRET) between full-length apoE4 labeled with EDANS at positions 62, 139, 210, 247, and 276 and tetramethylrhodamine (TMR)-labeled Aβ further support involvement of multiple regions of apoE in the interactions. Since the interactions involve intermediates of Aβ quantitative evaluation of the binding affinities are not feasible. Hence we employed a competitive binding assay to examine whether the N- and C-terminal domains interact cooperatively. Addition of unlabeled full-length apoE eliminates the FRET between EDANS-NTF + EDANS-CTF and TMR-Aβ almost completely but not vice versa. Furthermore, full-length apoE but not the equimolar mixture of the fragments could displace the already bound EDANS-apoE molecules from the complexes. Therefore, binding affinity of the Aβ oligomers to the intact full-length apoE is much higher than the affinity to the domains when mixed together as fragments. Thus, our results indicate that apoE-Aβ complex formation is mediated by positively cooperative multivalent binding between the multiple sites on apoE and the oligomeric forms of Aβ.
- Published
- 2018
- Full Text
- View/download PDF
50. Adaptive Multi-task Elastic Net based feature selection from Pharmacogenomics Databases
- Author
-
Souparno Ghosh, Ranadip Pall, Raziur Rahmanl, and Chamila Perera
- Subjects
0301 basic medicine ,Elastic net regularization ,Biological data ,Database ,Databases, Factual ,Computer science ,Feature extraction ,Feature selection ,computer.software_genre ,01 natural sciences ,Random forest ,Task (project management) ,010104 statistics & probability ,03 medical and health sciences ,030104 developmental biology ,Pharmacogenetics ,Pharmacogenomics ,Task analysis ,0101 mathematics ,computer ,Algorithms - Abstract
Integrating multiple databases of similar tasks is a significant problem in biological data analysis. In this paper, we consider whether feature selection in a single database can benefit from incorporating similar databases. We report that by using adaptive multi-task elastic net for feature selection and Random Forest for prediction, the prediction performance can be improved for pharmacogenomics databases. We also present a simulation study to explain the robust feature selection benefit of adaptive multi task elastic net while dealing with noisy features.
- Published
- 2018
Catalog
Discovery Service for Jio Institute Digital Library
For full access to our library's resources, please sign in.