Gabriel A. Preising, J. Faber-Hammond, S. Renn, Anna M. Ritz
ACM Reference Format: Gabriel A. Preising, Joshua J. Faber-Hammond, Suzy C. P. Renn, and Anna Ritz. 2020. A Protein-Protein Interactome for an African Cichlid. In Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics (BCB ’20), September 21–24, 2020, Virtual Event, USA. ACM, New York, NY, USA, 1 page. https://doi.org/10. 1145/3388440.3414916
ACM参考格式:Gabriel A. Preising, Joshua J. Faber-Hammond, Suzy C. P. Renn和Anna Ritz。2020。非洲慈鲷的蛋白质-蛋白质相互作用组。第11届ACM生物信息学、计算生物学和健康信息学国际会议论文集(BCB ' 20)， 2020年9月21-24日，虚拟事件，美国。ACM，纽约，美国，1页。https://doi.org/10。1145/3388440.3414916
{"title":"A Protein-Protein Interactome for an African Cichlid","authors":"Gabriel A. Preising, J. Faber-Hammond, S. Renn, Anna M. Ritz","doi":"10.1145/3388440.3414916","DOIUrl":"https://doi.org/10.1145/3388440.3414916","url":null,"abstract":"ACM Reference Format: Gabriel A. Preising, Joshua J. Faber-Hammond, Suzy C. P. Renn, and Anna Ritz. 2020. A Protein-Protein Interactome for an African Cichlid. In Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics (BCB ’20), September 21–24, 2020, Virtual Event, USA. ACM, New York, NY, USA, 1 page. https://doi.org/10. 1145/3388440.3414916","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"128 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123131206","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Horizontal gene transfer is one of the most important drivers of microbial gene and genome evolution. Despite its central role in microbial evolution, several aspects of horizontal gene transfer remain poorly understood. In particular, transfers can be either additive or replacing depending on whether the transferred gene adds itself as a new gene in the recipient genome or replaces an existing homologous gene. However, despite recent efforts, there do not yet exist effective computational approaches for classifying inferred transfers as being additive or replacing. In this work, we address this gap by devising a novel supervised machine learning approach for classifying transfers as being either additive or replacing. Our approach is based on phylogenetic reconciliation, a standard computational technique for inferring transfers. Our classifier, named ARTra, uses as features the classifications provided by several simple reconciliation-based classification rules, along with topological information from the gene tree, and ensembles them to produce a more accurate classification. ARTra is efficient and robust and significantly improves upon the classification accuracy of the only existing computational approach for this problem. We demonstrate the accuracy of ARTra by applying it to a wide range of simulated datasets and to a large biological dataset. Our results show that ARTra performs well over a broad range of evolutionary conditions and on real data, and that it does so even when trained only on a narrow range of such conditions and only using simulated data. An open-source implementation of ARTra is freely available from https://compbio.engr.uconn.edu/software/ARTra/.
{"title":"A Supervised Machine Learning Approach for Distinguishing Between Additive and Replacing Horizontal Gene Transfers","authors":"Abhijit Mondal, Misagh Kordi, Mukul S. Bansal","doi":"10.1145/3388440.3412428","DOIUrl":"https://doi.org/10.1145/3388440.3412428","url":null,"abstract":"Horizontal gene transfer is one of the most important drivers of microbial gene and genome evolution. Despite its central role in microbial evolution, several aspects of horizontal gene transfer remain poorly understood. In particular, transfers can be either additive or replacing depending on whether the transferred gene adds itself as a new gene in the recipient genome or replaces an existing homologous gene. However, despite recent efforts, there do not yet exist effective computational approaches for classifying inferred transfers as being additive or replacing. In this work, we address this gap by devising a novel supervised machine learning approach for classifying transfers as being either additive or replacing. Our approach is based on phylogenetic reconciliation, a standard computational technique for inferring transfers. Our classifier, named ARTra, uses as features the classifications provided by several simple reconciliation-based classification rules, along with topological information from the gene tree, and ensembles them to produce a more accurate classification. ARTra is efficient and robust and significantly improves upon the classification accuracy of the only existing computational approach for this problem. We demonstrate the accuracy of ARTra by applying it to a wide range of simulated datasets and to a large biological dataset. Our results show that ARTra performs well over a broad range of evolutionary conditions and on real data, and that it does so even when trained only on a narrow range of such conditions and only using simulated data. An open-source implementation of ARTra is freely available from https://compbio.engr.uconn.edu/software/ARTra/.","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"54 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124516747","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Fatemeh Rouzbeh, A. Grama, Paul M. Griffin, Mohammad Adibuzzaman
Improving healthcare depends on collecting and analyzing different types of health related data such as Electronic Health Records (EHR), Patient Generated Health Data (PGHD), prescription and medication data and medical image data. Even though different solutions in terms of storage and processing have been designed and developed but each solution is usually designed for a specific type of data. Storing, processing, and analyzing all types of data using a single solution necessarily doesn't result in best performance and quality of analysis. To acquire the better quality, each types of data requires its own type of storage, data processing and machine learning solutions which cannot be integrated as a unified system in some cases. In order to have a unified system that serves all types of data we propose a modular cloud native architecture with autonomous modules in terms of control, deployment and management for each types of data.
{"title":"A Unified Cloud-Native Architecture For Heterogeneous Data Aggregation And Computation","authors":"Fatemeh Rouzbeh, A. Grama, Paul M. Griffin, Mohammad Adibuzzaman","doi":"10.1145/3388440.3414911","DOIUrl":"https://doi.org/10.1145/3388440.3414911","url":null,"abstract":"Improving healthcare depends on collecting and analyzing different types of health related data such as Electronic Health Records (EHR), Patient Generated Health Data (PGHD), prescription and medication data and medical image data. Even though different solutions in terms of storage and processing have been designed and developed but each solution is usually designed for a specific type of data. Storing, processing, and analyzing all types of data using a single solution necessarily doesn't result in best performance and quality of analysis. To acquire the better quality, each types of data requires its own type of storage, data processing and machine learning solutions which cannot be integrated as a unified system in some cases. In order to have a unified system that serves all types of data we propose a modular cloud native architecture with autonomous modules in terms of control, deployment and management for each types of data.","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"62 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131206453","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Graphery","authors":"Heyuan Zeng, Anna M. Ritz","doi":"10.1145/3388440.3414915","DOIUrl":"https://doi.org/10.1145/3388440.3414915","url":null,"abstract":"","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116900559","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Tim Kosfeld, Jonathan McMillan, R. DiPaolo, Jie Hou, Tae-Hyuk Ahn
The adaptive immune system expresses millions of different receptors that detect and fight pathogens encountered throughout life. These receptors are encoded by unique DNA sequences that allow immune cells to express millions of different receptors. High-throughput sequencing and analyses of immune cell receptor sequences present a unique opportunity to inform our understanding of immunological responses to infections and to evaluate vaccine efficacy. Even after the infection is eliminated, pathogen-specific immune cells and their receptor sequences are present at higher frequencies than prior to infection, and their increase in frequency prevents secondary infections. As a result of their persistence in the body, they may be useful for diagnosing infections and evaluating vaccine efficacy as a stable biomarker. However, this process requires thorough analysis of massive datasets at an accuracy beyond traditional statistical tests to diagnose infectious statuses based on sequence analyses. Here we evaluate various machine learning and deep learning algorithms to measure the performance of the identification and diagnosis of specific viral infections or vaccination statuses using the publicly available mouse (monkeypox infection and smallpox vaccination) and human (cytomegalovirus serostatus) T-cell receptor sequenced datasets. Our intensive experiments hold the potential for effective screening of disease status, including recently encountered strains like the ongoing SARS-CoV-2 pandemic.
{"title":"Performance Evaluation of Viral Infection Diagnosis using T-Cell Receptor Sequence and Artificial Intelligence","authors":"Tim Kosfeld, Jonathan McMillan, R. DiPaolo, Jie Hou, Tae-Hyuk Ahn","doi":"10.1145/3388440.3412420","DOIUrl":"https://doi.org/10.1145/3388440.3412420","url":null,"abstract":"The adaptive immune system expresses millions of different receptors that detect and fight pathogens encountered throughout life. These receptors are encoded by unique DNA sequences that allow immune cells to express millions of different receptors. High-throughput sequencing and analyses of immune cell receptor sequences present a unique opportunity to inform our understanding of immunological responses to infections and to evaluate vaccine efficacy. Even after the infection is eliminated, pathogen-specific immune cells and their receptor sequences are present at higher frequencies than prior to infection, and their increase in frequency prevents secondary infections. As a result of their persistence in the body, they may be useful for diagnosing infections and evaluating vaccine efficacy as a stable biomarker. However, this process requires thorough analysis of massive datasets at an accuracy beyond traditional statistical tests to diagnose infectious statuses based on sequence analyses. Here we evaluate various machine learning and deep learning algorithms to measure the performance of the identification and diagnosis of specific viral infections or vaccination statuses using the publicly available mouse (monkeypox infection and smallpox vaccination) and human (cytomegalovirus serostatus) T-cell receptor sequenced datasets. Our intensive experiments hold the potential for effective screening of disease status, including recently encountered strains like the ongoing SARS-CoV-2 pandemic.","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117037691","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
David Perez, Ling-Hong Hung, Sonia Xu, K. Y. Yeung, W. Lloyd
Public Infrastructure-as-a-Service (IaaS) clouds abstract various details regarding the implementation of resources provided to users. For example, users are not informed about the exact physical location of their virtual machines (VMs), the specific hardware used, the number of co-resident VMs they reside with, or the workloads that co-resident VMs are running. Detecting when VMs underperform can help identify resource contention from co-resident VMs to spur their replacement. Resource utilization metrics can be used to help classify performance of runs for use in VM performance model datasets to sample the distribution of performance outcomes in the cloud. VM performance models are key to predicting the cost of bioinformatics analyses in the public cloud. This paper investigates the performance variations of running a RNA sequencing workflow in the public cloud. We examine causes of performance variations including VM provisioning, CPU heterogeneity, and resource contention. We leverage Amazon Elastic Compute Cloud (EC2) placement groups, a feature designed to help influence VM placement to help examine how VM placement impacts performance variations. As a use case, we investigate the performance of a multi-stage bioinformatics RNA sequencing (RNA-seq) analytical workflow consisting of four distinct phases, executing in 90 minutes on average using 8-core public cloud VMs. In addition, we investigate whether Linux resource utilization metrics collected by profiling workflow runs can help identify performance implications.
{"title":"An Investigation on Public Cloud Performance Variation for an RNA Sequencing Workflow","authors":"David Perez, Ling-Hong Hung, Sonia Xu, K. Y. Yeung, W. Lloyd","doi":"10.1145/3388440.3414859","DOIUrl":"https://doi.org/10.1145/3388440.3414859","url":null,"abstract":"Public Infrastructure-as-a-Service (IaaS) clouds abstract various details regarding the implementation of resources provided to users. For example, users are not informed about the exact physical location of their virtual machines (VMs), the specific hardware used, the number of co-resident VMs they reside with, or the workloads that co-resident VMs are running. Detecting when VMs underperform can help identify resource contention from co-resident VMs to spur their replacement. Resource utilization metrics can be used to help classify performance of runs for use in VM performance model datasets to sample the distribution of performance outcomes in the cloud. VM performance models are key to predicting the cost of bioinformatics analyses in the public cloud. This paper investigates the performance variations of running a RNA sequencing workflow in the public cloud. We examine causes of performance variations including VM provisioning, CPU heterogeneity, and resource contention. We leverage Amazon Elastic Compute Cloud (EC2) placement groups, a feature designed to help influence VM placement to help examine how VM placement impacts performance variations. As a use case, we investigate the performance of a multi-stage bioinformatics RNA sequencing (RNA-seq) analytical workflow consisting of four distinct phases, executing in 90 minutes on average using 8-core public cloud VMs. In addition, we investigate whether Linux resource utilization metrics collected by profiling workflow runs can help identify performance implications.","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132167520","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Empowered by the availability of high-performance computing (HPC) infrastructure (e.g. GPUs and HPC clusters), machine learning and optimization have become key technologies to analyze big bioinformatics data. In this keynote talk, I will present our machine learning and optimization algorithms for addressing three important data-intensive bioinformatics problems: (1) predicting protein tertiary structures from the evolutionary information in big protein sequence data generated by genomics and meta-genomics sequencing; (2) reconstructing high-resolution 3D genome conformations for integrating omics data; and (3) modeling gene regulatory networks from transcriptomics and genomics data, leveraging the high-performance computing platform available at the University of Missouri -- Columbia. The three research topics are briefly described below. Protein structure modeling on big protein sequence data. Predicting protein tertiary structure from sequence is a major challenge in bioinformatics and protein science. After a long period stagnancy, the field is experiencing a revolution driven by applying deep learning to leverage the amino acid (residue) evolutionary information hidden in the large amount of protein sequence data generated by the genome and meta-genome sequencing effort. In this talk, I will describe our deep convolutional neural network methods for predicting residue-residue contacts (e.g. interactions) and the distance-based method of reconstructing protein tertiary structures from predicted contacts that was ranked among the top methods in the 13th Critical Assessment of Techniques for Protein Structure Prediction in 2018 [1], along with Google DeepMind's AlphaFold. Reconstructing high-resolution 3D Conformations of large genomes for omics data analysis. 3D conformations (or structures) of genomes provide critical gene-gene and gene-enhancer interactions not available in 1D genome sequences. Unlike genome sequencing, there is no experimental technique to directly determine the 3D structure of genome. In this talk, I will present our high-performance, large-scale, data-driven optimization algorithm for reconstructing high-resolution 3D genome structures from deep chromosome conformation capturing (i.e. Hi-C) data [2]. The algorithm is highly scalable and efficient to reconstruct the 3D structures of large genomes such as the human genome at 5KB resolution. The high-resolution 3D genome models can be used to study gene function, gene expression, genome methylation and integrate multiple sources of omics data. Gene regulatory network modeling on transcriptomics and genomics data. Inferring gene regulatory relationships from large-scale gene expression data is an important, yet unsolved problem in bioinformatics. Gene regulatory networks provide a concise and informative representation of complex gene regulatory relationships. In this talk, I will present our probabilistic graphical model method for reliably reconstructing gene regulatory ne
{"title":"Large-Scale Machine Learning and Optimization for Bioinformatics Data Analysis","authors":"Jianlin Cheng","doi":"10.1145/3388440.3415587","DOIUrl":"https://doi.org/10.1145/3388440.3415587","url":null,"abstract":"Empowered by the availability of high-performance computing (HPC) infrastructure (e.g. GPUs and HPC clusters), machine learning and optimization have become key technologies to analyze big bioinformatics data. In this keynote talk, I will present our machine learning and optimization algorithms for addressing three important data-intensive bioinformatics problems: (1) predicting protein tertiary structures from the evolutionary information in big protein sequence data generated by genomics and meta-genomics sequencing; (2) reconstructing high-resolution 3D genome conformations for integrating omics data; and (3) modeling gene regulatory networks from transcriptomics and genomics data, leveraging the high-performance computing platform available at the University of Missouri -- Columbia. The three research topics are briefly described below. Protein structure modeling on big protein sequence data. Predicting protein tertiary structure from sequence is a major challenge in bioinformatics and protein science. After a long period stagnancy, the field is experiencing a revolution driven by applying deep learning to leverage the amino acid (residue) evolutionary information hidden in the large amount of protein sequence data generated by the genome and meta-genome sequencing effort. In this talk, I will describe our deep convolutional neural network methods for predicting residue-residue contacts (e.g. interactions) and the distance-based method of reconstructing protein tertiary structures from predicted contacts that was ranked among the top methods in the 13th Critical Assessment of Techniques for Protein Structure Prediction in 2018 [1], along with Google DeepMind's AlphaFold. Reconstructing high-resolution 3D Conformations of large genomes for omics data analysis. 3D conformations (or structures) of genomes provide critical gene-gene and gene-enhancer interactions not available in 1D genome sequences. Unlike genome sequencing, there is no experimental technique to directly determine the 3D structure of genome. In this talk, I will present our high-performance, large-scale, data-driven optimization algorithm for reconstructing high-resolution 3D genome structures from deep chromosome conformation capturing (i.e. Hi-C) data [2]. The algorithm is highly scalable and efficient to reconstruct the 3D structures of large genomes such as the human genome at 5KB resolution. The high-resolution 3D genome models can be used to study gene function, gene expression, genome methylation and integrate multiple sources of omics data. Gene regulatory network modeling on transcriptomics and genomics data. Inferring gene regulatory relationships from large-scale gene expression data is an important, yet unsolved problem in bioinformatics. Gene regulatory networks provide a concise and informative representation of complex gene regulatory relationships. In this talk, I will present our probabilistic graphical model method for reliably reconstructing gene regulatory ne","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"4 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126918998","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the world in the midst of a global pandemic, it is important to be able to quickly generate new drug-like compounds for drug research purposes. While some successful work has been done [3, 6] there is still much work to be done, especially as viruses like Coronavirus are notoriously hard to treat. Since peptide drugs are generally better at blocking protein-protein interactions than small molecule drugs [5], something important in anti-viral work, we will use our GANDALF methodology to generate new peptides to interact with targets of interest. Here we are working with two important COVID-19 targets: the SARS-CoV-2 main protease (M[Pro]) and the andangiotensin-converting enzyme 2 (ACE2). Covid-19 is able to enter human cells via interaction between its spike protein and ACE2 and, once in the cell, MPro breaks down polyproteins to create more of the virus [1]. We have generated peptides for each of our targets using our GANDALF (Generative Adversarial Network Drug-tArget Ligand Fructifier) methodology [4]. We compare our generated peptides with a previously discovered novel ACE2 inhibitor [2]. We also compare our results for MPro with a recently publish small molecule based on α-ketoamide inhibitors recently developed as a drug lead [7]. Our best generated peptide for ACE2 is a small, six residue peptide [SSNATV]. This peptide has a binding affinity of --29.880. The novel, peptide inhibitor previously designed has a binding affinity of -19.843. Our generated peptide has a lower binding affinity, which is generally more desirable and indicates more stable binding. However, the novel inhibitor is larger at 26 peptides and may be more suitable for use without the need for too many additional modifications. Our peptide though is a good starting place for further improvements and optimization. The binding affinity for our best generated peptide of MPro is --41.038. This peptide has a size of eleven residues [WWTWTPFHLLV]. Our peptide has a similar binding affinity to that of the small molecule, α-ketoamide based inhibitor is --5.501. Not only does our peptide have a better binding affinity, but as a peptide, it has the added advantage of being better able to disrupt the activity of the MPro than the small molecule inhibitor. It is also encouraging that our binding affinity for our best MPro generated peptide is comparable to the best available compounds. Peptide based drugs are an important part of viral treatment. Our work here provides reasonable starting peptides for further drug research and development.
{"title":"Novel Generated Peptides for COVID-19 Targets","authors":"Allison M. Rossetto, Wenjin Zhou","doi":"10.1145/3388440.3414919","DOIUrl":"https://doi.org/10.1145/3388440.3414919","url":null,"abstract":"With the world in the midst of a global pandemic, it is important to be able to quickly generate new drug-like compounds for drug research purposes. While some successful work has been done [3, 6] there is still much work to be done, especially as viruses like Coronavirus are notoriously hard to treat. Since peptide drugs are generally better at blocking protein-protein interactions than small molecule drugs [5], something important in anti-viral work, we will use our GANDALF methodology to generate new peptides to interact with targets of interest. Here we are working with two important COVID-19 targets: the SARS-CoV-2 main protease (M[Pro]) and the andangiotensin-converting enzyme 2 (ACE2). Covid-19 is able to enter human cells via interaction between its spike protein and ACE2 and, once in the cell, MPro breaks down polyproteins to create more of the virus [1]. We have generated peptides for each of our targets using our GANDALF (Generative Adversarial Network Drug-tArget Ligand Fructifier) methodology [4]. We compare our generated peptides with a previously discovered novel ACE2 inhibitor [2]. We also compare our results for MPro with a recently publish small molecule based on α-ketoamide inhibitors recently developed as a drug lead [7]. Our best generated peptide for ACE2 is a small, six residue peptide [SSNATV]. This peptide has a binding affinity of --29.880. The novel, peptide inhibitor previously designed has a binding affinity of -19.843. Our generated peptide has a lower binding affinity, which is generally more desirable and indicates more stable binding. However, the novel inhibitor is larger at 26 peptides and may be more suitable for use without the need for too many additional modifications. Our peptide though is a good starting place for further improvements and optimization. The binding affinity for our best generated peptide of MPro is --41.038. This peptide has a size of eleven residues [WWTWTPFHLLV]. Our peptide has a similar binding affinity to that of the small molecule, α-ketoamide based inhibitor is --5.501. Not only does our peptide have a better binding affinity, but as a peptide, it has the added advantage of being better able to disrupt the activity of the MPro than the small molecule inhibitor. It is also encouraging that our binding affinity for our best MPro generated peptide is comparable to the best available compounds. Peptide based drugs are an important part of viral treatment. Our work here provides reasonable starting peptides for further drug research and development.","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"120 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126305789","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Many real-world Electronic Health Record (EHR) data contain a large proportion of missing values. Leaving a substantial portion of missing information unaddressed usually causes significant bias, leading to invalid conclusions to be drawn. On the other hand, training a machine learning model with a much smaller nearly-complete subset can drastically impact the reliability and accuracy of model inference. Data imputation algorithms that attempt to replace missing data with meaningful values, inevitably increase the variability of effect estimates with increased missingness, making it unreliable for hypothesis validation. We propose a novel Ensemble-Learning for Missing Value (ELMV) framework, an effective approach to construct multiple subsets with much lower missing rates of the original EHR data as well as to mobilize dedicated support data for ensemble learning, for the purpose of reducing the bias caused by substantial missing values. ELMV has been evaluated on real-world healthcare data for critical feature identification and simulation data with different missing rates for outcome prediction. In both experiments, ELMV outperforms conventional missing value imputation methods and traditional ensemble learning models. The source code of ELMV is available at https://github.com/lucasliu0928/ELMV.
{"title":"ELMV","authors":"L. J. Liu, Hongwei Zhang, Jianzhong Di, Jin Chen","doi":"10.1145/3388440.3412431","DOIUrl":"https://doi.org/10.1145/3388440.3412431","url":null,"abstract":"Many real-world Electronic Health Record (EHR) data contain a large proportion of missing values. Leaving a substantial portion of missing information unaddressed usually causes significant bias, leading to invalid conclusions to be drawn. On the other hand, training a machine learning model with a much smaller nearly-complete subset can drastically impact the reliability and accuracy of model inference. Data imputation algorithms that attempt to replace missing data with meaningful values, inevitably increase the variability of effect estimates with increased missingness, making it unreliable for hypothesis validation. We propose a novel Ensemble-Learning for Missing Value (ELMV) framework, an effective approach to construct multiple subsets with much lower missing rates of the original EHR data as well as to mobilize dedicated support data for ensemble learning, for the purpose of reducing the bias caused by substantial missing values. ELMV has been evaluated on real-world healthcare data for critical feature identification and simulation data with different missing rates for outcome prediction. In both experiments, ELMV outperforms conventional missing value imputation methods and traditional ensemble learning models. The source code of ELMV is available at https://github.com/lucasliu0928/ELMV.","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"141 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124922879","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
B-cell epitope prediction was first developed to help design peptide-based vaccines for protective antibody-mediated immunity exemplified by neutralization of biological activity (e.g., pathogen infectivity). Requisite computational tools are benchmarked using experimentally obtained paratope-epitope binding data, which also serve as training data for machine-learning approaches to development of said tools. Such data are curated in the Immune Epitope Database (IEDB). However, IEDB curation guidelines define B-cell epitopes primarily on the basis of paratope-bound epitope structures, obscuring the crucial role of conformational disorder in the underlying immune recognition process. For the present work, pertinent IEDB B-cell assay records were retrieved and analyzed in relation to other data from both IEDB and external sources including the Protein Data Bank (PDB) and published literature, with special attention to data on conformational disorder among B-cell epitopes. This revealed examples of antipeptide antibodies that recognize conformationally disordered B-cell epitopes and thereby neutralize the biological activity of cognate targets (e.g., proteins and pathogens), with inconsistency noted in the definition of some epitopes. These results suggest an alternative approach to curating paratope-epitope binding data based on neutralization of biological activity by polyclonal antipeptide antibodies, with reference to immunogenic peptide sequences and their conformational disorder in the unbound state.
{"title":"Beyond B-Cell Epitopes: Curating Positive Data on Antipeptide Paratope Binding to Support Development of Computational Tools for Vaccine Design and Other Translational Applications","authors":"S. Caoili","doi":"10.1145/3388440.3414923","DOIUrl":"https://doi.org/10.1145/3388440.3414923","url":null,"abstract":"B-cell epitope prediction was first developed to help design peptide-based vaccines for protective antibody-mediated immunity exemplified by neutralization of biological activity (e.g., pathogen infectivity). Requisite computational tools are benchmarked using experimentally obtained paratope-epitope binding data, which also serve as training data for machine-learning approaches to development of said tools. Such data are curated in the Immune Epitope Database (IEDB). However, IEDB curation guidelines define B-cell epitopes primarily on the basis of paratope-bound epitope structures, obscuring the crucial role of conformational disorder in the underlying immune recognition process. For the present work, pertinent IEDB B-cell assay records were retrieved and analyzed in relation to other data from both IEDB and external sources including the Protein Data Bank (PDB) and published literature, with special attention to data on conformational disorder among B-cell epitopes. This revealed examples of antipeptide antibodies that recognize conformationally disordered B-cell epitopes and thereby neutralize the biological activity of cognate targets (e.g., proteins and pathogens), with inconsistency noted in the definition of some epitopes. These results suggest an alternative approach to curating paratope-epitope binding data based on neutralization of biological activity by polyclonal antipeptide antibodies, with reference to immunogenic peptide sequences and their conformational disorder in the unbound state.","PeriodicalId":411338,"journal":{"name":"Proceedings of the 11th ACM International Conference on Bioinformatics, Computational Biology and Health Informatics","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2020-09-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125205720","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：