Nagakishore Jammula, Sriram P. Chockalingam, S. Aluru
State-of-the-art high-throughput sequencing instruments decipher in excess of a billion short genomic fragments per run. The output sequences are referred to as 'reads'. These read datasets facilitate a wide variety of analyses with applications in areas such as genomics, metagenomics, and transcriptomics. Owing to the large size of the read datasets, such analyses are often compute and memory intensive. In this paper, we present a parallel algorithm for partitioning large-scale read datasets in order to facilitate distributed-memory parallel analyses. During the process of partitioning the read datasets, we construct and partition the associated de Bruijn graph in parallel. This allows applications that make use of a variant of the de Bruijn graph, such as de novo assembly, to directly leverage the generated de Bruijn graph partitions. In addition, we propose a mechanism for evaluating the quality of the generated partitions of reads and demonstrate that our algorithm produces high quality partitions. Our implementation is available at github.com/ParBLiSS/read_partitioning.
最先进的高通量测序仪器破译超过十亿短基因组片段每运行。输出序列被称为“读取”。这些读取数据集有助于在基因组学、宏基因组学和转录组学等领域进行各种分析。由于读取数据集的规模很大，这种分析通常需要大量的计算和内存。在本文中，我们提出了一种用于划分大规模读数据集的并行算法，以促进分布式内存并行分析。在划分读数据集的过程中，我们并行地构造和划分相关联的de Bruijn图。这允许使用de Bruijn图的变体的应用程序，例如de novo assembly，直接利用生成的de Bruijn图分区。此外，我们提出了一种评估读取分区质量的机制，并证明了我们的算法产生了高质量的分区。我们的实现可以在github.com/ParBLiSS/read_partitioning上获得。
{"title":"Distributed Memory Partitioning of High-Throughput Sequencing Datasets for Enabling Parallel Genomics Analyses","authors":"Nagakishore Jammula, Sriram P. Chockalingam, S. Aluru","doi":"10.1145/3107411.3107491","DOIUrl":"https://doi.org/10.1145/3107411.3107491","url":null,"abstract":"State-of-the-art high-throughput sequencing instruments decipher in excess of a billion short genomic fragments per run. The output sequences are referred to as 'reads'. These read datasets facilitate a wide variety of analyses with applications in areas such as genomics, metagenomics, and transcriptomics. Owing to the large size of the read datasets, such analyses are often compute and memory intensive. In this paper, we present a parallel algorithm for partitioning large-scale read datasets in order to facilitate distributed-memory parallel analyses. During the process of partitioning the read datasets, we construct and partition the associated de Bruijn graph in parallel. This allows applications that make use of a variant of the de Bruijn graph, such as de novo assembly, to directly leverage the generated de Bruijn graph partitions. In addition, we propose a mechanism for evaluating the quality of the generated partitions of reads and demonstrate that our algorithm produces high quality partitions. Our implementation is available at github.com/ParBLiSS/read_partitioning.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115698458","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}
Edward W. Huang, Sheng Wang, Bingxue Li, Ran Zhang, Baoyan Liu, Runshun Zhang, Jie Liu, Xuezhong Zhou, Hongsheng Lin, ChengXiang Zhai
The continual growth of electronic medical record (EMR) databases has paved the way for many data mining applications, including the discovery of novel disease-drug associations and the prediction of patient survival rates. However, these tasks are hindered because EMRs are usually segmented or incomplete. EMR analysis is further limited by the overabundance of medical term synonyms and morphologies, which causes existing techniques to mismatch records containing semantically similar but lexically distinct terms. Current solutions fill in missing values with techniques that tend to introduce noise rather than reduce it. In this paper, we propose to simultaneously infer missing data and solve semantic mismatching in EMRs by first integrating EMR data with molecular interaction networks and domain knowledge to build the HEMnet, a heterogeneous medical information network. We then project this network onto a low-dimensional space, and group entities in the network according to their relative distances. Lastly, we use this entity distance information to enrich the original EMRs. We evaluate the effectiveness of this method according to its ability to separate patients with dissimilar survival functions. We show that our method can obtain significant (p-value < 0.01) results for each cancer subtype in a lung cancer dataset, while the baselines cannot.
{"title":"HEMnet: Integration of Electronic Medical Records with Molecular Interaction Networks and Domain Knowledge for Survival Analysis","authors":"Edward W. Huang, Sheng Wang, Bingxue Li, Ran Zhang, Baoyan Liu, Runshun Zhang, Jie Liu, Xuezhong Zhou, Hongsheng Lin, ChengXiang Zhai","doi":"10.1145/3107411.3107422","DOIUrl":"https://doi.org/10.1145/3107411.3107422","url":null,"abstract":"The continual growth of electronic medical record (EMR) databases has paved the way for many data mining applications, including the discovery of novel disease-drug associations and the prediction of patient survival rates. However, these tasks are hindered because EMRs are usually segmented or incomplete. EMR analysis is further limited by the overabundance of medical term synonyms and morphologies, which causes existing techniques to mismatch records containing semantically similar but lexically distinct terms. Current solutions fill in missing values with techniques that tend to introduce noise rather than reduce it. In this paper, we propose to simultaneously infer missing data and solve semantic mismatching in EMRs by first integrating EMR data with molecular interaction networks and domain knowledge to build the HEMnet, a heterogeneous medical information network. We then project this network onto a low-dimensional space, and group entities in the network according to their relative distances. Lastly, we use this entity distance information to enrich the original EMRs. We evaluate the effectiveness of this method according to its ability to separate patients with dissimilar survival functions. We show that our method can obtain significant (p-value < 0.01) results for each cancer subtype in a lung cancer dataset, while the baselines cannot.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114572817","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}
Virtually all molecular interactions networks, independent of organism and physiological context, have majority-leaves minority-hubs (mLmH) topology. Current generative models of this topology are based on controversial hypotheses that, controversy aside, demonstrate sufficient but not necessary evolutionary conditions for its emergence. Here we show that the circumvention of computational intractability provides sufficient and (assuming P!=NP) necessary conditions for the emergence of the mLmH property. Evolutionary pressure on molecular interaction networks is simulated by randomly labelling some interactions as 'beneficial' and others 'detrimental'. Each gene is subsequently given a benefit (damage) score according to how many beneficial (detrimental) interactions it is projecting onto or attracting from other genes. The problem of identifying which subset of genes should ideally be conserved and which deleted, so as to maximize (minimize) the total number of beneficial (detrimental) interactions network-wide, is NP-hard. An evolutionary algorithm that simulates hypothetical instances of this problem and selects for networks that produce the easiest instances leads to networks that possess the mLmH property. The degree distributions of synthetically evolved networks match those of publicly available experimentally-validated biological networks from many phylogenetically-distant organisms.
{"title":"Computational Intractability Generates the Topology of Biological Networks","authors":"Ali A Atiia, Corbin Hopper, J. Waldispühl","doi":"10.1145/3107411.3107453","DOIUrl":"https://doi.org/10.1145/3107411.3107453","url":null,"abstract":"Virtually all molecular interactions networks, independent of organism and physiological context, have majority-leaves minority-hubs (mLmH) topology. Current generative models of this topology are based on controversial hypotheses that, controversy aside, demonstrate sufficient but not necessary evolutionary conditions for its emergence. Here we show that the circumvention of computational intractability provides sufficient and (assuming P!=NP) necessary conditions for the emergence of the mLmH property. Evolutionary pressure on molecular interaction networks is simulated by randomly labelling some interactions as 'beneficial' and others 'detrimental'. Each gene is subsequently given a benefit (damage) score according to how many beneficial (detrimental) interactions it is projecting onto or attracting from other genes. The problem of identifying which subset of genes should ideally be conserved and which deleted, so as to maximize (minimize) the total number of beneficial (detrimental) interactions network-wide, is NP-hard. An evolutionary algorithm that simulates hypothetical instances of this problem and selects for networks that produce the easiest instances leads to networks that possess the mLmH property. The degree distributions of synthetically evolved networks match those of publicly available experimentally-validated biological networks from many phylogenetically-distant organisms.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114590476","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}
Mansooreh Ahmadian, Shuo Wang, J. Tyson, Young Cao
The budding yeast cell cycle is regulated by complex and multi-scale control mechanisms, and is subject to inherent noise in a cell, resulted from low copy numbers of species such as critical mRNAs. Conventional deterministic models cannot capture this inherent noise. Although stochastic models can generate simulation results to better represent inherent noise in system dynamics, the stochastic approach is often computationally too expensive for complex systems, which exhibit multiscale features in two aspects: species with different scales of abundances and reactions with different scales of firing frequencies. To address this challenge, one promising solution is to adopt a hybrid approach. It replaces the single mathematical representation of either discrete-stochastic formulation or continuous deterministic formulation with an integration of both methods, so that the corresponding advantageous features in both methods are well kept to achieve a trade-off between accuracy and efficiency. In this work, we propose a hybrid stochastic model that represents the regulatory network of the budding yeast cell cycle control mechanism, respectively, by Gillespie's stochastic simulation algorithm (SSA) and ordinary differential equations (ODEs). Simulation results of our model were compared with published experimental measurement on the budding yeast cell cycle. The comparison demonstrates that our hybrid model well represents many critical characteristics of the budding yeast cell cycle, and reproduces more than 100 phenotypes of mutant cases. Moreover, the model accounts for partial viability of certain mutant strains. The last but not the least, the proposed scheme is shown to be considerably faster in both modeling and simulation than the equivalent stochastic simulation.
{"title":"Hybrid ODE/SSA Model of the Budding Yeast Cell Cycle Control Mechanism with Mutant Case Study","authors":"Mansooreh Ahmadian, Shuo Wang, J. Tyson, Young Cao","doi":"10.1145/3107411.3107437","DOIUrl":"https://doi.org/10.1145/3107411.3107437","url":null,"abstract":"The budding yeast cell cycle is regulated by complex and multi-scale control mechanisms, and is subject to inherent noise in a cell, resulted from low copy numbers of species such as critical mRNAs. Conventional deterministic models cannot capture this inherent noise. Although stochastic models can generate simulation results to better represent inherent noise in system dynamics, the stochastic approach is often computationally too expensive for complex systems, which exhibit multiscale features in two aspects: species with different scales of abundances and reactions with different scales of firing frequencies. To address this challenge, one promising solution is to adopt a hybrid approach. It replaces the single mathematical representation of either discrete-stochastic formulation or continuous deterministic formulation with an integration of both methods, so that the corresponding advantageous features in both methods are well kept to achieve a trade-off between accuracy and efficiency. In this work, we propose a hybrid stochastic model that represents the regulatory network of the budding yeast cell cycle control mechanism, respectively, by Gillespie's stochastic simulation algorithm (SSA) and ordinary differential equations (ODEs). Simulation results of our model were compared with published experimental measurement on the budding yeast cell cycle. The comparison demonstrates that our hybrid model well represents many critical characteristics of the budding yeast cell cycle, and reproduces more than 100 phenotypes of mutant cases. Moreover, the model accounts for partial viability of certain mutant strains. The last but not the least, the proposed scheme is shown to be considerably faster in both modeling and simulation than the equivalent stochastic simulation.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125359846","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}
Discriminating native-like complexes from false-positives with high accuracy is one of the biggest challenges in protein-protein docking. The relationship between various favorable intermolecular interactions (e.g., Van der Waals, electrostatic, desolvation forces, etc.) and the similarity of a conformation to its native structure is commonly agreed, though the precise nature of this relationship is not known very well. Existing protein-protein docking methods typically formulate this relationship as a weighted sum of selected terms and tune their weights by introducing a training set with which they evaluate and rank candidate complexes. Despite improvements in recent docking methods, they are still producing a large number of false positives, which often leads to incorrect prediction of complex binding. Using machine learning, we implemented an approach that not only ranks candidate complexes relative to each other, but also predicts how similar each candidate is to the native conformation. We built a Support Vector Regressor (SVR) using physico-chemical features and evolutionary conservation. We trained and tested the model on extensive datasets of complexes generated by three state-of-the-art docking methods. The set of docked complexes was generated from 79 different protein-protein complexes in both the rigid and medium categories of the Protein-Protein Docking Benchmark v.5. We were able to generally outperform the built-in scoring functions of the docking programs we used to generate the complexes, attesting to the potential of our approach in predicting the correct binding of protein-protein complexes.
{"title":"Ranking Protein-Protein Binding Using Evolutionary Information and Machine Learning","authors":"R. Farhoodi, Bahar Akbal-Delibas, Nurit Haspel","doi":"10.1145/3107411.3107497","DOIUrl":"https://doi.org/10.1145/3107411.3107497","url":null,"abstract":"Discriminating native-like complexes from false-positives with high accuracy is one of the biggest challenges in protein-protein docking. The relationship between various favorable intermolecular interactions (e.g., Van der Waals, electrostatic, desolvation forces, etc.) and the similarity of a conformation to its native structure is commonly agreed, though the precise nature of this relationship is not known very well. Existing protein-protein docking methods typically formulate this relationship as a weighted sum of selected terms and tune their weights by introducing a training set with which they evaluate and rank candidate complexes. Despite improvements in recent docking methods, they are still producing a large number of false positives, which often leads to incorrect prediction of complex binding. Using machine learning, we implemented an approach that not only ranks candidate complexes relative to each other, but also predicts how similar each candidate is to the native conformation. We built a Support Vector Regressor (SVR) using physico-chemical features and evolutionary conservation. We trained and tested the model on extensive datasets of complexes generated by three state-of-the-art docking methods. The set of docked complexes was generated from 79 different protein-protein complexes in both the rigid and medium categories of the Protein-Protein Docking Benchmark v.5. We were able to generally outperform the built-in scoring functions of the docking programs we used to generate the complexes, attesting to the potential of our approach in predicting the correct binding of protein-protein complexes.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"79 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127713111","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}
Alevtina Dubovitskaya, T. Buclin, M. Schumacher, K. Aberer, Y. Thoma
Therapeutic Drug Monitoring (TDM) is a key concept in precision medicine. The goal of TDM is to avoid therapeutic failure or toxic effects of a drug due to insufficient or excessive circulating concentration exposure related to between-patient variability in the drug's disposition. We present TUCUXI - an intelligent system for TDM. By making use of embedded mathematical models, the software allows to compute maximum likelihood individual predictions of drug concentrations from population pharmacokinetic data, based on patient's parameters and previously observed concentrations. TUCUXI was developed to be used in medical practice, to assist clinicians in taking dosage adjustment decisions for optimizing drug concentration levels. This software is currently being tested in a University Hospital. In this paper we focus on the process of software integration in clinical workflow. The modular architecture of the software allows us to plug in a module enabling data aggregation for research purposes. This is an important feature in order to develop new mathematical models for drugs, and thus to improve TDM. Finally we discuss ethical issues related to the use of an automated decision support system in clinical practice, in particular if it allows data aggregation for research purposes.
{"title":"TUCUXI: An Intelligent System for Personalized Medicine from Individualization of Treatments to Research Databases and Back","authors":"Alevtina Dubovitskaya, T. Buclin, M. Schumacher, K. Aberer, Y. Thoma","doi":"10.1145/3107411.3107439","DOIUrl":"https://doi.org/10.1145/3107411.3107439","url":null,"abstract":"Therapeutic Drug Monitoring (TDM) is a key concept in precision medicine. The goal of TDM is to avoid therapeutic failure or toxic effects of a drug due to insufficient or excessive circulating concentration exposure related to between-patient variability in the drug's disposition. We present TUCUXI - an intelligent system for TDM. By making use of embedded mathematical models, the software allows to compute maximum likelihood individual predictions of drug concentrations from population pharmacokinetic data, based on patient's parameters and previously observed concentrations. TUCUXI was developed to be used in medical practice, to assist clinicians in taking dosage adjustment decisions for optimizing drug concentration levels. This software is currently being tested in a University Hospital. In this paper we focus on the process of software integration in clinical workflow. The modular architecture of the software allows us to plug in a module enabling data aggregation for research purposes. This is an important feature in order to develop new mathematical models for drugs, and thus to improve TDM. Finally we discuss ethical issues related to the use of an automated decision support system in clinical practice, in particular if it allows data aggregation for research purposes.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124299298","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}
Sriram P. Chockalingam, M. Aluru, Hongqing Guo, Yanhai Yin, S. Aluru
Motivation: Reverse engineering gene networks from expression data is a widelymstudied problem, for which numerous mathematical models have been developed. Network reconstruction methods can be used to study specific pathways, or can be applied at the whole-genome scale to analyze large compendiums of expression datasets to uncover genome-wide interactions. However, few methods can scale to such large number of genes and experiments, and to date, genome-scale comparative assessment of network reconstruction methods has largely been limited to simpler organisms such as E. coli. Results: In this paper, we analyze 11,760 microarray experiments on the model plant Arabidopsis thaliana drawn from public repositories. We generate genome scale networks of Arabidopsis using three different methods -- Pearson correlation, mutual information, and graphical Gaussian modeling -- and analyze and compare these networks to test for their robustness in successfully recovering relationships between functionally related genes. We demonstrate that functional grouping of microarray experiments into different tissue types and experimental conditions is important to discover context-specific interactions. Our comparisons include benchmarking against experimentally confirmed interactions, the Arabidopsis network resource AraNet, and study of specific pathways. Our results show that networks generated by the mutual information based method have better characteristics in terms of functional modularity as measured by both connected component and sub-network extraction analysis with respect to gene sets selected from brassinosteroid and stress regulation pathways. Availability: The classification datasets and constructed genome-scale networks are publicly available at the location http://alurulab.cc.gatech.edu/arabidopsis-networks
{"title":"Reverse Engineering Gene Networks: A Comparative Study at Genome-scale","authors":"Sriram P. Chockalingam, M. Aluru, Hongqing Guo, Yanhai Yin, S. Aluru","doi":"10.1145/3107411.3107428","DOIUrl":"https://doi.org/10.1145/3107411.3107428","url":null,"abstract":"Motivation: Reverse engineering gene networks from expression data is a widelymstudied problem, for which numerous mathematical models have been developed. Network reconstruction methods can be used to study specific pathways, or can be applied at the whole-genome scale to analyze large compendiums of expression datasets to uncover genome-wide interactions. However, few methods can scale to such large number of genes and experiments, and to date, genome-scale comparative assessment of network reconstruction methods has largely been limited to simpler organisms such as E. coli. Results: In this paper, we analyze 11,760 microarray experiments on the model plant Arabidopsis thaliana drawn from public repositories. We generate genome scale networks of Arabidopsis using three different methods -- Pearson correlation, mutual information, and graphical Gaussian modeling -- and analyze and compare these networks to test for their robustness in successfully recovering relationships between functionally related genes. We demonstrate that functional grouping of microarray experiments into different tissue types and experimental conditions is important to discover context-specific interactions. Our comparisons include benchmarking against experimentally confirmed interactions, the Arabidopsis network resource AraNet, and study of specific pathways. Our results show that networks generated by the mutual information based method have better characteristics in terms of functional modularity as measured by both connected component and sub-network extraction analysis with respect to gene sets selected from brassinosteroid and stress regulation pathways. Availability: The classification datasets and constructed genome-scale networks are publicly available at the location http://alurulab.cc.gatech.edu/arabidopsis-networks","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123630849","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}
Studying the breast cancer survival genes information will help to enhance the treatment and save more patents life by identifying the genes biomarker to recommend the proper treatment type. That is why it is now a great challenge for researchers to have more research on breast cancer specially with the great enhancement in the fields of bioinformatics, data mining, and machine learning techniques which were a new revolution in the cancer treatment. A dataset contains the survival information and treatments methods for 1980 female breast cancer patient is used for building the prediction model, the gene expression are the features of the learning model [1], where the combination of the survival and treatments information are the classes. A hierarchal model that consists of hybrid feature selection and classification method are utilized to differentiate a class from the rest of the classes. The results show that a few number of gene biomarkers (gene signature) at each node which can determine the class with accuracy around 99% for survival living / deceased based on treatments which is vital to ensure that the patients will have the best potential response to a specific therapy. This signatures will be used as a predictor of survival in breast cancer.
{"title":"Machine Learning Model for Identifying Gene Biomarkers for Breast Cancer Treatment Survival","authors":"A. Tabl, A. Alkhateeb, W. ElMaraghy, A. Ngom","doi":"10.1145/3107411.3108217","DOIUrl":"https://doi.org/10.1145/3107411.3108217","url":null,"abstract":"Studying the breast cancer survival genes information will help to enhance the treatment and save more patents life by identifying the genes biomarker to recommend the proper treatment type. That is why it is now a great challenge for researchers to have more research on breast cancer specially with the great enhancement in the fields of bioinformatics, data mining, and machine learning techniques which were a new revolution in the cancer treatment. A dataset contains the survival information and treatments methods for 1980 female breast cancer patient is used for building the prediction model, the gene expression are the features of the learning model [1], where the combination of the survival and treatments information are the classes. A hierarchal model that consists of hybrid feature selection and classification method are utilized to differentiate a class from the rest of the classes. The results show that a few number of gene biomarkers (gene signature) at each node which can determine the class with accuracy around 99% for survival living / deceased based on treatments which is vital to ensure that the patients will have the best potential response to a specific therapy. This signatures will be used as a predictor of survival in breast cancer.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"569 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122103769","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}
Probabilistic topic modeling is an active research field in machine learning and has been mainly used as an analytical tool to structure large textual corpora for data mining. It offers a viable approach to structure huge textual document collections into latent topic themes to aid text mining. Latent Dirichlet Allocation (LDA) is the most commonly used topic modelling method across a wide number of technical fields. However, model development can be arduous and tedious, and requires burdensome and systematic sensitivity studies in order to find the best set of model parameters. In this study, we use a heuristic approach to estimate the most appropriate number of topics. Specifically, the rate of perplexity change (RPC) as a function of numbers of topics is proposed as a suitable selector. We test the stability and effectiveness of the proposed method for three markedly different types of grounded-truth datasets: Salmonella next generation sequencing, pharmacological side effects, and textual abstracts on computational biology and bioinformatics (TCBB) from PubMed. Then we describe extensive sensitivity studies to determine best practices for generating effective topic models. To test effectiveness and validity of topic models, we constructed a ground truth data set from PubMed that contained some 40 health related themes including negative controls, and mixed it with a data set of unstructured documents. We found that obtaining the most useful model, tuned to desired sensitivity versus specificity, requires an iterative process wherein preprocessing steps, the type of topic modeling algorithm, and the algorithm's model parameters are systematically varied. Models need to be compared with both qualitative, subjective assessments and quantitative, objective assessments, and care is required that Gibbs sampling in model estimation is sufficient to assure stable solutions. With a high quality model, documents can be rank-ordered in accordance with probability of being associated with complex regulatory query string, greatly lessoning text mining work. Importantly, topic models are agnostic about how words and documents are defined, and thus our findings are extensible to topic models where samples are defined as documents, and genes, proteins or their sequences are words.
{"title":"Best Setting of Model Parameters in Applying Topic Modeling on Textual Documents.","authors":"Wen Zou, Weizhong Zhao, James J. Chen, R. Perkins","doi":"10.1145/3107411.3108195","DOIUrl":"https://doi.org/10.1145/3107411.3108195","url":null,"abstract":"Probabilistic topic modeling is an active research field in machine learning and has been mainly used as an analytical tool to structure large textual corpora for data mining. It offers a viable approach to structure huge textual document collections into latent topic themes to aid text mining. Latent Dirichlet Allocation (LDA) is the most commonly used topic modelling method across a wide number of technical fields. However, model development can be arduous and tedious, and requires burdensome and systematic sensitivity studies in order to find the best set of model parameters. In this study, we use a heuristic approach to estimate the most appropriate number of topics. Specifically, the rate of perplexity change (RPC) as a function of numbers of topics is proposed as a suitable selector. We test the stability and effectiveness of the proposed method for three markedly different types of grounded-truth datasets: Salmonella next generation sequencing, pharmacological side effects, and textual abstracts on computational biology and bioinformatics (TCBB) from PubMed. Then we describe extensive sensitivity studies to determine best practices for generating effective topic models. To test effectiveness and validity of topic models, we constructed a ground truth data set from PubMed that contained some 40 health related themes including negative controls, and mixed it with a data set of unstructured documents. We found that obtaining the most useful model, tuned to desired sensitivity versus specificity, requires an iterative process wherein preprocessing steps, the type of topic modeling algorithm, and the algorithm's model parameters are systematically varied. Models need to be compared with both qualitative, subjective assessments and quantitative, objective assessments, and care is required that Gibbs sampling in model estimation is sufficient to assure stable solutions. With a high quality model, documents can be rank-ordered in accordance with probability of being associated with complex regulatory query string, greatly lessoning text mining work. Importantly, topic models are agnostic about how words and documents are defined, and thus our findings are extensible to topic models where samples are defined as documents, and genes, proteins or their sequences are words.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"18 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125678791","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}
Recombination is an evolutionary force that results in mosaic genomes for microorganisms. The evolutionary history of microorganisms cannot be properly inferred if recombination has occurred among a set of taxa. That is, polymorphic sites of a multiple sequence alignment cannot be described by a single phylogenetic tree. Thus, detecting the presence of recombination is crucial before phylogeny inference. The phylogenetic-based methods are commonly utilized to explore recombination, however, the compatibility-based methods are more computationally efficient since the phylogeny construction is not required. We propose a novel approach focusing on the pairwise compatibility of polymorphic sites of given regions to characterize potential breakpoints in recombinant bacterial and viral genomes. The performance of average compatibility ratio (ACR) approach is evaluated on simulated alignments of different scenarios comparing with two programs, GARD and RDP4. Three empirical datasets of varying genome sizes with varying levels of homoplasy are also utilized for testing. The results demonstrate that our approach is able to detect the presence of recombination and identify the recombinant breakpoints efficiently, which provides a better understanding of distinct phylogenies among mosaic sequences.
{"title":"A Compatibility Approach to Identify Recombination Breakpoints in Bacterial and Viral Genomes","authors":"Yi-Pin Lai, T. Ioerger","doi":"10.1145/3107411.3107432","DOIUrl":"https://doi.org/10.1145/3107411.3107432","url":null,"abstract":"Recombination is an evolutionary force that results in mosaic genomes for microorganisms. The evolutionary history of microorganisms cannot be properly inferred if recombination has occurred among a set of taxa. That is, polymorphic sites of a multiple sequence alignment cannot be described by a single phylogenetic tree. Thus, detecting the presence of recombination is crucial before phylogeny inference. The phylogenetic-based methods are commonly utilized to explore recombination, however, the compatibility-based methods are more computationally efficient since the phylogeny construction is not required. We propose a novel approach focusing on the pairwise compatibility of polymorphic sites of given regions to characterize potential breakpoints in recombinant bacterial and viral genomes. The performance of average compatibility ratio (ACR) approach is evaluated on simulated alignments of different scenarios comparing with two programs, GARD and RDP4. Three empirical datasets of varying genome sizes with varying levels of homoplasy are also utilized for testing. The results demonstrate that our approach is able to detect the presence of recombination and identify the recombinant breakpoints efficiently, which provides a better understanding of distinct phylogenies among mosaic sequences.","PeriodicalId":246388,"journal":{"name":"Proceedings of the 8th ACM International Conference on Bioinformatics, Computational Biology,and Health Informatics","volume":"18 3","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-08-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131672164","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：