Data Scientist @ BBVA Spain, previously at Experian UK and Royal Holloway, University of London. Supervisors: Luis M. de Campos and Juan M. Fernández-Luna
We introduce a MeSH-based method that accurately quantifies similarity between heritable diseases... more We introduce a MeSH-based method that accurately quantifies similarity between heritable diseases at molecular level. This method effectively brings together the existing information about diseases that is scattered across the vast corpus of biomedical literature. We prove that sets of MeSH terms provide a highly descriptive representation of heritable disease and that the structure of MeSH provides a natural way of combining individual MeSH vocabularies. We show that our measure can be used effectively in the prediction of candidate disease genes. We developed a web application to query more than 28.5 million relationships between 7,574 hereditary diseases (96% of OMIM) based on our similarity measure.
Objective
In the context of “network medicine”, gene prioritization methods represent one of the... more Objective
In the context of “network medicine”, gene prioritization methods represent one of the main tools to discover candidate disease genes by exploiting the large amount of data covering different types of functional relationships between genes. Several works proposed to integrate multiple sources of data to improve disease gene prioritization, but to our knowledge no systematic studies focused on the quantitative evaluation of the impact of network integration on gene prioritization. In this paper, we aim at providing an extensive analysis of gene-disease associations not limited to genetic disorders, and a systematic comparison of different network integration methods for gene prioritization.
Materials and methods
We collected nine different functional networks representing different functional relationships between genes, and we combined them through both unweighted and weighted network integration methods. We then prioritized genes with respect to each of the considered 708 medical subject headings (MeSH) diseases by applying classical guilt-by-association, random walk and random walk with restart algorithms, and the recently proposed kernelized score functions.
Results
The results obtained with classical random walk algorithms and the best single network achieved an average area under the curve (AUC) across the 708 MeSH diseases of about 0.82, while kernelized score functions and network integration boosted the average AUC to about 0.89. Weighted integration, by exploiting the different “informativeness” embedded in different functional networks, outperforms unweighted integration at 0.01 significance level, according to the Wilcoxon signed rank sum test. For each MeSH disease we provide the top-ranked unannotated candidate genes, available for further bio-medical investigation.
Conclusions
Network integration is necessary to boost the performances of gene prioritization methods. Moreover the methods based on kernelized score functions can further enhance disease gene ranking results, by adopting both local and global learning strategies, able to exploit the overall topology of the network.
Summary: We present GOssTo, the Gene Ontology semantic similarity Tool, a software system for cal... more Summary: We present GOssTo, the Gene Ontology semantic similarity Tool, a software system for calculating semantic similarities between gene products according to the Gene Ontology. GOssTo is bundled with six semantic similarity measures, including both term-based and graph-based measures, and has extension capabilities to allow the user to add new similarities. Importantly, for any measure, GOssTo can also calculate the Random Walk Contribution (Yang et al., 2012) that has been shown to greatly improve the accuracy of similarity measures. GOssTo is very fast, easy to use, and it allows the calculation of similarities on a genomic scale in a few minutes on a regular desktop machine.
Availability: GOssTo is available both as a stand-alone application running on GNU/Linux, Windows and MacOS from www.paccanarolab.org/gossto and as a web application from www.paccanarolab.org/gosstoweb. The stand-alone application features a simple and concise command line interface for easy integration into high-throughput data processing pipelines.
We propose a method which, given a document to be classified, automatically generates an ordered ... more We propose a method which, given a document to be classified, automatically generates an ordered set of appropriate descriptors extracted from a thesaurus. The method creates a Bayesian network to model the thesaurus and uses probabilistic inference to select the set of descriptors having high posterior probability of being relevant given the available evidence (the document to be classified). Our model can be used without having preclassified training documents, although it improves its performance as long as more training data become available. We have tested the classification model using a document dataset containing parliamentary resolutions from the regional Parliament of Andalucía at Spain, which were manually indexed from the Eurovoc thesaurus, also carrying out an experimental comparison with other standard text classifiers.
Information Retrieval (IR) is a leading research field nowadays around the world, because the eff... more Information Retrieval (IR) is a leading research field nowadays around the world, because the effective and efficient access to the great amount of information resources available in the Internet, among other sources, is a real problem. From an educational point of view, this fact has caused an increasing presence of IR courses in postgraduate studies. In general, we can say that the contents of these courses depend on the convenience of the research groups working on IR in each particular University.
This is the third year that researchers from the University of Granada (specifically from the Unc... more This is the third year that researchers from the University of Granada (specifically from the Uncertainty Treatment in Artificial Intelligence research group) participate on the XML Document Mining Track of the INEX workshop. As in previous editions, we restrict our solutions to the application of probabilistic methods to these problems. To be more precise, we are looking to solve the problem of link-based document classification within the field of Bayesian networks [10] (a special case of probabilistic graphical models).
Focused Access to XML Documents, 6th International Workshop of the Initiative for the Evaluation of XML Retrieval, INEX 2007, Dagstuhl Castle, Germany, December 17-19, 2007. Selected Papers. Springer 2008 Lecture Notes in Computer Science ISBN 978-3-540-85901-7, Dec 2007
This paper exposes the results of our participation at INEX’07 in the AdHoc track and the compari... more This paper exposes the results of our participation at INEX’07 in the AdHoc track and the comparison of these results with respect to the ones obtained last year. Three runs were submitted to each of the Focused, Relevant In Context and Best In Context tasks,all of them obtained with Garnata, our Information Retrieval Systemfor structured documents. As in the past year, we use a model based on Influence Diagrams, the CID model. The result of our participation has been better than the last year and we have reached an acceptable position in the ranking for the three tasks. In the paper we describe the model, the system and we show the differences between our systems at INEX’06 and INEX’07, which make possible to get a better performance
Symbolic and Quantitative Approaches to Reasoning with Uncertainty, 9th European Conference, ECSQARU 2007, Hammamet, Tunisia, October 31 - November 2, 2007, Proceedings. Springer 2007 Lecture Notes in Computer Science ISBN 978-3-540-75255-4: 865-877, 2007
"
We propose a method which, given a document to be classified, automatically generates an ord... more "
We propose a method which, given a document to be classified, automatically generates an ordered set of appropriate descriptors extracted from a thesaurus. The method creates a Bayesian network to model the thesaurus and uses probabilistic inference to select the set of descriptors having high posterior probability of being relevant given the available evidence (the document to be classified). We apply the method to the classification of parliamentary initiatives in the regional Parliament of Andalucía at Spain from the Eurovoc thesaurus.
"
Focused Access to XML Documents, 6th International Workshop of the Initiative for the Evaluation of XML Retrieval, INEX 2007, Dagstuhl Castle, Germany, December 17-19, 2007. Selected Papers. Springer 2008 Lecture Notes in Computer Science ISBN 978-3-540-85901-7, Dec 2007
This paper exposes the results of our participation in the
Document Mining track at INEX’07. We ... more This paper exposes the results of our participation in the
Document Mining track at INEX’07. We have focused on the task of classification of XML documents. Our approach to deal with structured document representations uses classification methods for plain text, applied to flattened versions of the documents, where some of their structural properties have been translated to plain text. We have explored several options to convert structured documents into flat documents, in combination with two probabilistic methods for text categorization. The main conclusion of our experiments is that taking advantage of document structure to improve classification results is a difficult task.
Comparative Evaluation of XML Information Retrieval Systems, 5th International Workshop of the Initiative for the Evaluation of XML Retrieval, INEX 2006, Dagstuhl Castle, Germany, December 17-20, 2006, Springer 2007 Lecture Notes in Computer Science ISBN 978-3-540-73887-9, Dec 2006
This paper exposes the results of our participation in INEX’06. Two runs were submitted to the Ad... more This paper exposes the results of our participation in INEX’06. Two runs were submitted to the Ad Hoc Thorough track obtained with Garnata, our Information Retrieval system for structured documents. We have implemented two different models based on Influence Diagrams, the SID and CID models. The result of this first participation has been very poor. In the paper, we describe the models, the system, and analyse the possible reason of such a bad performance.
We introduce a MeSH-based method that accurately quantifies similarity between heritable diseases... more We introduce a MeSH-based method that accurately quantifies similarity between heritable diseases at molecular level. This method effectively brings together the existing information about diseases that is scattered across the vast corpus of biomedical literature. We prove that sets of MeSH terms provide a highly descriptive representation of heritable disease and that the structure of MeSH provides a natural way of combining individual MeSH vocabularies. We show that our measure can be used effectively in the prediction of candidate disease genes. We developed a web application to query more than 28.5 million relationships between 7,574 hereditary diseases (96% of OMIM) based on our similarity measure.
Objective
In the context of “network medicine”, gene prioritization methods represent one of the... more Objective
In the context of “network medicine”, gene prioritization methods represent one of the main tools to discover candidate disease genes by exploiting the large amount of data covering different types of functional relationships between genes. Several works proposed to integrate multiple sources of data to improve disease gene prioritization, but to our knowledge no systematic studies focused on the quantitative evaluation of the impact of network integration on gene prioritization. In this paper, we aim at providing an extensive analysis of gene-disease associations not limited to genetic disorders, and a systematic comparison of different network integration methods for gene prioritization.
Materials and methods
We collected nine different functional networks representing different functional relationships between genes, and we combined them through both unweighted and weighted network integration methods. We then prioritized genes with respect to each of the considered 708 medical subject headings (MeSH) diseases by applying classical guilt-by-association, random walk and random walk with restart algorithms, and the recently proposed kernelized score functions.
Results
The results obtained with classical random walk algorithms and the best single network achieved an average area under the curve (AUC) across the 708 MeSH diseases of about 0.82, while kernelized score functions and network integration boosted the average AUC to about 0.89. Weighted integration, by exploiting the different “informativeness” embedded in different functional networks, outperforms unweighted integration at 0.01 significance level, according to the Wilcoxon signed rank sum test. For each MeSH disease we provide the top-ranked unannotated candidate genes, available for further bio-medical investigation.
Conclusions
Network integration is necessary to boost the performances of gene prioritization methods. Moreover the methods based on kernelized score functions can further enhance disease gene ranking results, by adopting both local and global learning strategies, able to exploit the overall topology of the network.
Summary: We present GOssTo, the Gene Ontology semantic similarity Tool, a software system for cal... more Summary: We present GOssTo, the Gene Ontology semantic similarity Tool, a software system for calculating semantic similarities between gene products according to the Gene Ontology. GOssTo is bundled with six semantic similarity measures, including both term-based and graph-based measures, and has extension capabilities to allow the user to add new similarities. Importantly, for any measure, GOssTo can also calculate the Random Walk Contribution (Yang et al., 2012) that has been shown to greatly improve the accuracy of similarity measures. GOssTo is very fast, easy to use, and it allows the calculation of similarities on a genomic scale in a few minutes on a regular desktop machine.
Availability: GOssTo is available both as a stand-alone application running on GNU/Linux, Windows and MacOS from www.paccanarolab.org/gossto and as a web application from www.paccanarolab.org/gosstoweb. The stand-alone application features a simple and concise command line interface for easy integration into high-throughput data processing pipelines.
We propose a method which, given a document to be classified, automatically generates an ordered ... more We propose a method which, given a document to be classified, automatically generates an ordered set of appropriate descriptors extracted from a thesaurus. The method creates a Bayesian network to model the thesaurus and uses probabilistic inference to select the set of descriptors having high posterior probability of being relevant given the available evidence (the document to be classified). Our model can be used without having preclassified training documents, although it improves its performance as long as more training data become available. We have tested the classification model using a document dataset containing parliamentary resolutions from the regional Parliament of Andalucía at Spain, which were manually indexed from the Eurovoc thesaurus, also carrying out an experimental comparison with other standard text classifiers.
Information Retrieval (IR) is a leading research field nowadays around the world, because the eff... more Information Retrieval (IR) is a leading research field nowadays around the world, because the effective and efficient access to the great amount of information resources available in the Internet, among other sources, is a real problem. From an educational point of view, this fact has caused an increasing presence of IR courses in postgraduate studies. In general, we can say that the contents of these courses depend on the convenience of the research groups working on IR in each particular University.
This is the third year that researchers from the University of Granada (specifically from the Unc... more This is the third year that researchers from the University of Granada (specifically from the Uncertainty Treatment in Artificial Intelligence research group) participate on the XML Document Mining Track of the INEX workshop. As in previous editions, we restrict our solutions to the application of probabilistic methods to these problems. To be more precise, we are looking to solve the problem of link-based document classification within the field of Bayesian networks [10] (a special case of probabilistic graphical models).
Focused Access to XML Documents, 6th International Workshop of the Initiative for the Evaluation of XML Retrieval, INEX 2007, Dagstuhl Castle, Germany, December 17-19, 2007. Selected Papers. Springer 2008 Lecture Notes in Computer Science ISBN 978-3-540-85901-7, Dec 2007
This paper exposes the results of our participation at INEX’07 in the AdHoc track and the compari... more This paper exposes the results of our participation at INEX’07 in the AdHoc track and the comparison of these results with respect to the ones obtained last year. Three runs were submitted to each of the Focused, Relevant In Context and Best In Context tasks,all of them obtained with Garnata, our Information Retrieval Systemfor structured documents. As in the past year, we use a model based on Influence Diagrams, the CID model. The result of our participation has been better than the last year and we have reached an acceptable position in the ranking for the three tasks. In the paper we describe the model, the system and we show the differences between our systems at INEX’06 and INEX’07, which make possible to get a better performance
Symbolic and Quantitative Approaches to Reasoning with Uncertainty, 9th European Conference, ECSQARU 2007, Hammamet, Tunisia, October 31 - November 2, 2007, Proceedings. Springer 2007 Lecture Notes in Computer Science ISBN 978-3-540-75255-4: 865-877, 2007
"
We propose a method which, given a document to be classified, automatically generates an ord... more "
We propose a method which, given a document to be classified, automatically generates an ordered set of appropriate descriptors extracted from a thesaurus. The method creates a Bayesian network to model the thesaurus and uses probabilistic inference to select the set of descriptors having high posterior probability of being relevant given the available evidence (the document to be classified). We apply the method to the classification of parliamentary initiatives in the regional Parliament of Andalucía at Spain from the Eurovoc thesaurus.
"
Focused Access to XML Documents, 6th International Workshop of the Initiative for the Evaluation of XML Retrieval, INEX 2007, Dagstuhl Castle, Germany, December 17-19, 2007. Selected Papers. Springer 2008 Lecture Notes in Computer Science ISBN 978-3-540-85901-7, Dec 2007
This paper exposes the results of our participation in the
Document Mining track at INEX’07. We ... more This paper exposes the results of our participation in the
Document Mining track at INEX’07. We have focused on the task of classification of XML documents. Our approach to deal with structured document representations uses classification methods for plain text, applied to flattened versions of the documents, where some of their structural properties have been translated to plain text. We have explored several options to convert structured documents into flat documents, in combination with two probabilistic methods for text categorization. The main conclusion of our experiments is that taking advantage of document structure to improve classification results is a difficult task.
Comparative Evaluation of XML Information Retrieval Systems, 5th International Workshop of the Initiative for the Evaluation of XML Retrieval, INEX 2006, Dagstuhl Castle, Germany, December 17-20, 2006, Springer 2007 Lecture Notes in Computer Science ISBN 978-3-540-73887-9, Dec 2006
This paper exposes the results of our participation in INEX’06. Two runs were submitted to the Ad... more This paper exposes the results of our participation in INEX’06. Two runs were submitted to the Ad Hoc Thorough track obtained with Garnata, our Information Retrieval system for structured documents. We have implemented two different models based on Influence Diagrams, the SID and CID models. The result of this first participation has been very poor. In the paper, we describe the models, the system, and analyse the possible reason of such a bad performance.
Actas de las VI Jornadas de Transferencia Tecnológica de Inteligencia Artificial (TTIA'2005), 37-44, ISBN: 84-9732-435-8, 2005.
En este trabajo se presenta el prototipo de sistema de recuperación de información diseñado para ... more En este trabajo se presenta el prototipo de sistema de recuperación de información diseñado para permitir la recuperación de los diarios de sesiones del Parlamento de Andalucía. Este software está basado en el Context-based Influence Diagram Model, modelo de recuperación cuya base formal son las redes bayesianas y los diagramas de influencia, mo delos gráficos probabilísticos ampliamente utilizados en
el tratamiento de la incertidumbre.
Abstract: Multilabel classification is a relatively recent subfield of machine learning. Unlike t... more Abstract: Multilabel classification is a relatively recent subfield of machine learning. Unlike to the classical approach, where patterns are labeled with only one category, in multilabel classification, an arbitrary number of categories is chosen to label a pattern. Due to the problem complexity (the solution is one among an exponential number of alternatives), a very common solution (the binary method) is frequently used, learning a binary classifier for every category, and combining them all afterwards.
Abstract: In this paper, we propose the V-index (or, Virtuosity index) as a novel metric to asses... more Abstract: In this paper, we propose the V-index (or, Virtuosity index) as a novel metric to assess the scientific virtuosity of academics. This index can be applied to researchers and journals as well. In particular, we show that the V-index fills the gap of h-index and similar metrics in considering the self-citations of authors or journals.
Geometry and Information Retrieval Alfonso E. Romero aeromero@decsai.ugr.es ... The aim here is t... more Geometry and Information Retrieval Alfonso E. Romero aeromero@decsai.ugr.es ... The aim here is to define a probability measure over the set of sub-spaces of H, using the natural identification of each subspace with ... [6]R. Cooke, M. Keane and W. Moran, An elementary proof ...
Uploads
In the context of “network medicine”, gene prioritization methods represent one of the main tools to discover candidate disease genes by exploiting the large amount of data covering different types of functional relationships between genes. Several works proposed to integrate multiple sources of data to improve disease gene prioritization, but to our knowledge no systematic studies focused on the quantitative evaluation of the impact of network integration on gene prioritization. In this paper, we aim at providing an extensive analysis of gene-disease associations not limited to genetic disorders, and a systematic comparison of different network integration methods for gene prioritization.
Materials and methods
We collected nine different functional networks representing different functional relationships between genes, and we combined them through both unweighted and weighted network integration methods. We then prioritized genes with respect to each of the considered 708 medical subject headings (MeSH) diseases by applying classical guilt-by-association, random walk and random walk with restart algorithms, and the recently proposed kernelized score functions.
Results
The results obtained with classical random walk algorithms and the best single network achieved an average area under the curve (AUC) across the 708 MeSH diseases of about 0.82, while kernelized score functions and network integration boosted the average AUC to about 0.89. Weighted integration, by exploiting the different “informativeness” embedded in different functional networks, outperforms unweighted integration at 0.01 significance level, according to the Wilcoxon signed rank sum test. For each MeSH disease we provide the top-ranked unannotated candidate genes, available for further bio-medical investigation.
Conclusions
Network integration is necessary to boost the performances of gene prioritization methods. Moreover the methods based on kernelized score functions can further enhance disease gene ranking results, by adopting both local and global learning strategies, able to exploit the overall topology of the network.
Availability: GOssTo is available both as a stand-alone application running on GNU/Linux, Windows and MacOS from www.paccanarolab.org/gossto and as a web application from www.paccanarolab.org/gosstoweb. The stand-alone application features a simple and concise command line interface for easy integration into high-throughput data processing pipelines.
We propose a method which, given a document to be classified, automatically generates an ordered set of appropriate descriptors extracted from a thesaurus. The method creates a Bayesian network to model the thesaurus and uses probabilistic inference to select the set of descriptors having high posterior probability of being relevant given the available evidence (the document to be classified). We apply the method to the classification of parliamentary initiatives in the regional Parliament of Andalucía at Spain from the Eurovoc thesaurus.
"
Document Mining track at INEX’07. We have focused on the task of classification of XML documents. Our approach to deal with structured document representations uses classification methods for plain text, applied to flattened versions of the documents, where some of their structural properties have been translated to plain text. We have explored several options to convert structured documents into flat documents, in combination with two probabilistic methods for text categorization. The main conclusion of our experiments is that taking advantage of document structure to improve classification results is a difficult task.
In the context of “network medicine”, gene prioritization methods represent one of the main tools to discover candidate disease genes by exploiting the large amount of data covering different types of functional relationships between genes. Several works proposed to integrate multiple sources of data to improve disease gene prioritization, but to our knowledge no systematic studies focused on the quantitative evaluation of the impact of network integration on gene prioritization. In this paper, we aim at providing an extensive analysis of gene-disease associations not limited to genetic disorders, and a systematic comparison of different network integration methods for gene prioritization.
Materials and methods
We collected nine different functional networks representing different functional relationships between genes, and we combined them through both unweighted and weighted network integration methods. We then prioritized genes with respect to each of the considered 708 medical subject headings (MeSH) diseases by applying classical guilt-by-association, random walk and random walk with restart algorithms, and the recently proposed kernelized score functions.
Results
The results obtained with classical random walk algorithms and the best single network achieved an average area under the curve (AUC) across the 708 MeSH diseases of about 0.82, while kernelized score functions and network integration boosted the average AUC to about 0.89. Weighted integration, by exploiting the different “informativeness” embedded in different functional networks, outperforms unweighted integration at 0.01 significance level, according to the Wilcoxon signed rank sum test. For each MeSH disease we provide the top-ranked unannotated candidate genes, available for further bio-medical investigation.
Conclusions
Network integration is necessary to boost the performances of gene prioritization methods. Moreover the methods based on kernelized score functions can further enhance disease gene ranking results, by adopting both local and global learning strategies, able to exploit the overall topology of the network.
Availability: GOssTo is available both as a stand-alone application running on GNU/Linux, Windows and MacOS from www.paccanarolab.org/gossto and as a web application from www.paccanarolab.org/gosstoweb. The stand-alone application features a simple and concise command line interface for easy integration into high-throughput data processing pipelines.
We propose a method which, given a document to be classified, automatically generates an ordered set of appropriate descriptors extracted from a thesaurus. The method creates a Bayesian network to model the thesaurus and uses probabilistic inference to select the set of descriptors having high posterior probability of being relevant given the available evidence (the document to be classified). We apply the method to the classification of parliamentary initiatives in the regional Parliament of Andalucía at Spain from the Eurovoc thesaurus.
"
Document Mining track at INEX’07. We have focused on the task of classification of XML documents. Our approach to deal with structured document representations uses classification methods for plain text, applied to flattened versions of the documents, where some of their structural properties have been translated to plain text. We have explored several options to convert structured documents into flat documents, in combination with two probabilistic methods for text categorization. The main conclusion of our experiments is that taking advantage of document structure to improve classification results is a difficult task.
el tratamiento de la incertidumbre.