Applied text mining

This document discusses text mining and data integration techniques used to extract information from biomedical literature and databases. It describes named entity recognition to identify concepts, co-mentioning analysis to find associations between entities, and using these methods along with experimental data and predictions to build integrated networks of genes and proteins and their relationships. These networks are made accessible through web resources that unify data from various sources under common identifiers and provide visualization and programmatic access.

Text mining

Text mining techniques can be used to extract information and insights from the exponential growth of scientific literature. Key techniques include information retrieval to find relevant papers, named entity recognition to identify concepts, and information extraction to formalize facts. These techniques can be evaluated using benchmarking against manually annotated corpora, though creating such resources requires significant effort and the pragmatic approach of inspecting text mining outputs is much less work.

Literature mining: what is it, and should I care?

Mining literature and medical records

This document discusses mining literature and medical records using text mining techniques. It summarizes that text mining can be used to extract relevant information from large collections of scientific papers and medical records by using techniques like named entity recognition to identify concepts, information extraction to formalize stated facts, and analyzing co-mentioning of entities to find relationships. Challenges include the unstructured nature of medical records, differences between languages and formats, and privacy concerns when using patient health information. When applied carefully, text mining of literature and medical records can help identify new relationships and insights not captured in existing curated databases or help with medical research questions.

Literature mining and large-scale data integration

Biomedical literature mining

This document discusses various methodologies for extracting information from biological literature, including information retrieval, entity recognition, information extraction, and text/data mining. It provides an overview of different approaches like using co-occurrence, natural language processing, and machine learning methods. It also discusses challenges like integrating text with other data types and dealing with issues like ambiguity. Examples of existing text mining tools and their potential applications are also described.

Biomedical literature mining (and why we really need open access)

The document discusses various techniques for literature mining and systems biology including information retrieval, entity recognition, information extraction, text mining, and integration of text and biological data. It provides examples and status of different methods, from established techniques for information retrieval, entity recognition, and simple information extraction to improving advanced natural language processing-based information extraction and methods for text mining and integration of text and data.

Integration of biomedical literature and databases

Biological literature mining - from information retrieval to biological disco...

Text mining

Text mining can summarize scientific documents in 3 sentences or less by identifying key entities and relationships. It recognizes concepts like genes, proteins, diseases and extracts facts from text. This extracted information can then be integrated with other data to create more useful resources and provide novel insights through augmented browsing and analysis. Text mining aims to make navigating vast amounts of scientific literature simpler and less boring.

Exploring proteins, chemicals and their interactions with STRING and STITCH

biocs

This document summarizes databases and computational methods for exploring protein and chemical interactions. It introduces STRING and STITCH, which integrate various sources of data to predict interactions. STRING contains information on protein-protein interactions for 373 genomes. STITCH contains data on interactions between proteins and chemicals, including drugs. The document provides an example of using NetworKIN to predict kinase-substrate relationships and discusses how these databases and methods can provide insights into interaction networks and biological functions.

Biomedical text mining

This document discusses biomedical text mining techniques used to extract information from scientific papers. It covers named entity recognition to identify concepts like proteins, chemicals and diseases. It also discusses information extraction to formalize facts stated in text, such as interactions between biological components. Techniques include co-mentioning analysis and natural language processing and tools have been applied to large text corpora to aid discovery.

Open access - making the most of biomedical literature mining

An Introduction to Crispr Genome Editing

Chris Thorne

Theory and practice of graphical population analysis

Biology, Utrecht University

Hoofdstuk 16 2008 deel 1

The document discusses several key topics regarding the chromosomal basis of inheritance: 1) Early experiments transformed views on inheritance by showing bacterial cells could be transformed with DNA from other bacteria, and DNA was found to be the genetic material in viruses. 2) The structure of DNA was determined in the 1950s, with Rosalind Franklin's X-ray crystallography work revealing the double helix structure of DNA, which Watson and Crick modeled based on her data. 3) DNA replication involves separation of the DNA strands and synthesis of new complementary strands, with specific proteins involved in starting replication at the origin of replication site.

CRISPR Gene Editing Congress, 25-27 February 2015 in Boston, MA

Diane McKenna

Rewriting the Genome Using CRISPR and Synthetic Biology

Integrated DNA Technologies

The CRISPR/Cas9 system has emerged as one of the leading tools for modifying genomes of organisms ranging from E. coli to humans. Additionally, the simple gene targeting mechanism of CRISPR technology has been modified and adapted to other applications that include gene regulation, detection of intercellular trafficking, and pathogen detection. With a wealth of methods for introducing Cas9 and gRNAs into cells, it can be challenging to decide where to start. In this presentation, Dr Adam Clore describes the CRISPR mechanism and some of the most prominent uses for CRISPR, along with methods where IDT technologies can assist scientists in designing, testing, and executing a variety of CRISPR-mediated experiments. For more informaton, visit: http://www.idtdna.com/crispr

Mason abrf single_cell_2017

Christopher Mason

1. Single-cell RNA sequencing was performed on hematopoietic stem cells isolated from myelodysplastic syndrome patients and normal individuals to characterize heterogeneity. Cells were collected before and after treatment with decitabine from responders and non-responders. 2. Differential expression analysis identified genes dysregulated in MDS compared to normal, including pathways involved in hematopoiesis. Clusters of patients were identified based on expression of hematopoietic stem cell signature genes. 3. The study aims to understand heterogeneity in MDS, factors influencing response to therapy, and disease progression by characterizing gene expression profiles at the single-cell level. This may help identify new therapeutic targets.

Genome Editing Comes of Age

This document provides an overview of genome editing techniques such as CRISPR/Cas9 and rAAV and considerations for their use. It discusses how CRISPR/Cas9 and rAAV work to edit genomes and compares their advantages. Key factors for CRISPR gene editing are discussed such as gRNA design, donor design, and screening/validation approaches. The document also summarizes research optimizing CRISPR gene editing through improvements like testing different donor lengths and modifications. The goal is to translate genetic information into personalized medicines by leveraging tools like CRISPR and rAAV.

Hippocampal transcriptomic responses to technical and biological perturbations

Rayna Harris

This document summarizes a study investigating the effects of technical and biological perturbations on hippocampal gene expression profiles from single-cell transcriptomics. Specifically, it examines how the process of cellular dissociation alters gene expression and the ability to identify meaningful expression patterns. The study approaches this by analyzing up-regulated and down-regulated genes and pathways following dissociation. It also analyzes the effects of a stressful experience and cognitive training on hippocampal transcriptomes. The results reveal both unique and shared responses to technical and biological perturbations at the single-cell level through meta-analysis.

Sept2016 smallvar 10_x

GenomeInABottle

1) This document discusses 10X Genomics' linked read technology and its ability to improve genome assembly and variant calling compared to standard short read sequencing. 2) The linked read data from 10X Genomics can correctly place short reads into paralogous loci, improving genome alignment. This leads to improved variant calling compared to standard short read data. 3) Novel variants called from 10X Genomics data and validated using PacBio long reads showed higher validation rates compared to variants called from standard short read data. The document proposes further validation of the variant validation method.

Genome Editing Comes of Age; CRISPR, rAAV and the new landscape of molecular ...

Information is no longer a bottleneck, emphasis is shifting to the ‘what does it all mean’ In a translational context we hope that by answering that question we will be able to is to characterise the genetics that drive disease, and indeed develop drugs and diagnostics that are personalised to patients. Genome editing provides the link between the information here, and this outcome here, by allowing scientists to recapitulate specific genetic alterations in any gene in any living tissue to probe function, develop disease models and identify therapeutic strategies. So, not only do we now have unparalleled access to genetic information, but we now have the tools to most accuartely understand what this genetic information – with genome editing allowing us to explore the genetic drivers of disease in physiological models. AAV is a single-stranded, linear DNA virus with a a 4.7 kb genome which for the purpose of genome editing is replaced almost in entirety with the targeting vector sequence (except for the iTRs) It is in effect a highly effective DNA delivery mechanism After entry of the vector into the cell, target-specific homologous DNA is believed to activate and recruit HR-dependent repair factors can induce HR at rates approximately 1,000 times greater than plasmid based double stranded DNA vectors, but the mechanism by which it achieves this is still largely unknown By including a selection cassette can select for cells that have integrated the targeting vector, and then screen for clones which have undergone targeted insetion rather than random integration, which will generally be around 1%.

CRISPR presentation extended Mouse Modeling

Tristan Kempston

Genome engineering using CRISPR/Cas9 has several advantages over traditional gene targeting methods: it is faster, more precise, applicable to many species, and less expensive. CRISPR/Cas9 uses the Cas9 nuclease guided by a single guide RNA to introduce double-strand breaks at targeted genomic loci. This can generate gene knockouts through error-prone non-homologous end joining or allow for targeted insertions and modifications through homology-directed repair. While CRISPR/Cas9 has great potential, careful design of guide RNAs and donor templates is needed to minimize off-target effects.

Transhumanismo y Mejoramiento Genético mediante CRISPR

Bioeticared

This document discusses CRISPR gene editing technology, including its origins and development, applications for therapeutic purposes, and debates around its use. It describes how CRISPR uses a guide RNA and Cas9 protein to cut DNA in a targeted way. While promising for treating genetic diseases, some argue its use in human embryos or germline cells raises safety and ethical issues due to risks of uncontrolled errors passing to future generations. Transhumanists support modifying the germline to induce enhanced traits, but this remains controversial.

171017 giab for giab grc workshop

Systems biology: Large-scale biomedical data mining

The document discusses large-scale biomedical data mining and systems biology. It mentions several tools and databases for mining protein networks and interactions, including STRING, Reflect, NetworKIN, and Cyclebase. The tools use techniques like text mining, computational predictions, and experimental data to build dynamic, cell-specific models of signaling networks and understand systems properties over time.

Network biology: A basis for large-scale biomedical data mining

The document discusses network biology and large-scale data mining techniques used to analyze biomedical data. It describes several databases and tools developed including NetPhorest for predicting kinase-substrate relationships from sequence motifs, STRING for mapping protein-protein interaction networks across 630 genomes, and methods to predict drug side effects and potential new uses based on shared targets and side effect similarities. It also acknowledges contributions to developing these resources from researchers across several institutions.

What's hot

Literature Mining and Systems Biology

Integration of biomedical literature and databases

Biological literature mining - from information retrieval to biological disco...

Text mining

Exploring proteins, chemicals and their interactions with STRING and STITCH

biocs

Biomedical text mining

Open access - making the most of biomedical literature mining

An Introduction to Crispr Genome Editing

Chris Thorne

Theory and practice of graphical population analysis

Biology, Utrecht University

Hoofdstuk 16 2008 deel 1

CRISPR Gene Editing Congress, 25-27 February 2015 in Boston, MA

Diane McKenna

Rewriting the Genome Using CRISPR and Synthetic Biology

Integrated DNA Technologies

Mason abrf single_cell_2017

Christopher Mason

Genome Editing Comes of Age

Hippocampal transcriptomic responses to technical and biological perturbations

Rayna Harris

Sept2016 smallvar 10_x

GenomeInABottle

Genome Editing Comes of Age; CRISPR, rAAV and the new landscape of molecular ...

CRISPR presentation extended Mouse Modeling

Tristan Kempston

Transhumanismo y Mejoramiento Genético mediante CRISPR

Bioeticared

171017 giab for giab grc workshop

What's hot (20)

Literature Mining and Systems Biology

Integration of biomedical literature and databases

Biological literature mining - from information retrieval to biological disco...

Text mining

Exploring proteins, chemicals and their interactions with STRING and STITCH

Biomedical text mining

Open access - making the most of biomedical literature mining

An Introduction to Crispr Genome Editing

Theory and practice of graphical population analysis

Hoofdstuk 16 2008 deel 1

CRISPR Gene Editing Congress, 25-27 February 2015 in Boston, MA

Rewriting the Genome Using CRISPR and Synthetic Biology

Mason abrf single_cell_2017

Genome Editing Comes of Age

Hippocampal transcriptomic responses to technical and biological perturbations

Sept2016 smallvar 10_x

Genome Editing Comes of Age; CRISPR, rAAV and the new landscape of molecular ...

CRISPR presentation extended Mouse Modeling

Transhumanismo y Mejoramiento Genético mediante CRISPR

171017 giab for giab grc workshop

Viewers also liked

Systems biology: Large-scale biomedical data mining

Network biology: A basis for large-scale biomedical data mining

Medical data and text mining: Linking diseases, drugs, and adverse reactions

The document discusses medical data and text mining techniques used to link diseases, drugs, and adverse reactions using Danish healthcare registries. It describes how Danish registries contain data from central registries and individual hospitals dating back to 1968. These registries contain information on patients' diagnoses, medications, and hospital admissions. The document also discusses how natural language processing is used to extract structured information from unstructured free-text clinical notes to identify diseases, drugs, adverse reactions and relationships between them. This information can then be used to discover new adverse drug reactions and their frequencies.

Protein networks: A basis for large-scale data mining

The document discusses protein networks and how they can be used as a basis for large-scale data mining. It describes three phases: 1) constructing association networks using various data types, 2) building signaling networks by identifying signaling events and sequence motifs, and 3) examining dynamic networks to understand when interactions occur, such as during cell cycle regulation. The goal is to integrate different data sources to generate specific predictions and gain insights into systems properties and functions.

The pragmatic text miner: It's just another type of poorly standardized data

The document discusses text mining and its uses, including summarizing unstructured biomedical literature and integrating various data sources. It addresses challenges like different data formats, identifiers, and quality. Collaboration with domain experts is important to determine what topics to analyze and how best to solve problems. Text mining can help curate knowledge from experimental data, predictions, and online resources about proteins, chemicals, subcellular localization, tissue expression, and disease associations.

Data integration with STRING

The document discusses data integration strategies used by STRING, a database of known and predicted protein-protein interactions. STRING combines interaction data from numerous sources like experimental repositories, curated databases, text mining, and computational prediction methods. It maps identifiers across different databases, assigns confidence scores to interactions, and provides an interactive web interface for users to access and explore the integrated interaction network. External groups can contribute additional interaction data to STRING by hosting their own public data servers or providing data files in specified formats.

Using networks to derive function

This document discusses using networks to derive biological function from genomic data. It mentions several types of data that can be used like gene expression, protein-protein interactions, genetic interactions, pathways, literature mining, and co-mentioning in text. It also notes challenges integrating these diverse data sources that have different formats, identifiers, quality, and are spread across many databases and genomes. Lastly, it recommends combining all available evidence to predict functional associations.

One tagger, many uses - Illustrating the power of ontologies in named entity ...

The document describes a C++ tagger that can recognize named entities in biomedical literature with high precision and recall. It can identify molecular entities, genes, proteins, chemicals, and can assess studiedness, association networks, localization, expressions, tissues, diseases, side effects, organisms, and habitats. The tagger is fast, flexible, inherently thread-safe, and uses ontologies, dictionaries, expansion rules, and blacklists to identify entities. It has been used in various databases and tools for data integration, literature mining, and interactive annotation.

Viewers also liked (8)

Systems biology: Large-scale biomedical data mining

Network biology: A basis for large-scale biomedical data mining

Medical data and text mining: Linking diseases, drugs, and adverse reactions

Protein networks: A basis for large-scale data mining

The pragmatic text miner: It's just another type of poorly standardized data

Data integration with STRING

Using networks to derive function

One tagger, many uses - Illustrating the power of ontologies in named entity ...

Similar to Applied text mining

Computational approaches to cell cycle analysis: Current research topics (tho...

Integration of biomedical literature and databases

Open access - making the most of biomedical literature mining

STRING: Large-scale data and text mining

This document discusses large-scale data and text mining techniques used by STRING to build comprehensive protein association networks. STRING integrates information from genomic context, high-throughput experiments, co-expression and curated databases to assign a confidence score to each association. Natural language processing is applied to mine the scientific literature and extract entity and relation information from millions of articles and abstracts to expand the known protein association networks beyond curated knowledge. STRING is freely accessible online and allows users to perform queries and analyze networks for various organisms.

Text mining for protein and small molecule relations

The document discusses using text mining to identify relationships between proteins and small molecules mentioned in biomedical documents. It describes techniques for entity recognition and identification, as well as methods for extracting relationships between entities using co-occurrence analysis and natural language processing. Examples are provided to illustrate how relationships can be identified between proteins mentioned in a sample text passage.

Network biology: Large-scale data and text mining

This document discusses network biology and large-scale data and text mining. It describes how Lars Jensen uses computational predictions from over 1100 genomes along with experimental data and information extracted from text to build protein-protein association networks in STRING. These networks integrate known and predicted protein-protein interactions with functional associations, and are used to study biological systems at the network level.

Systems biology - Understanding biology at the systems level

The document discusses systems biology and its goal of understanding biology at the systems level. It explains that systems biology studies complete biological systems by integrating multiple types of high-throughput omics data and mathematical modeling. It provides examples of modeling the cell cycle and integrating gene expression, protein interaction, and genetic interaction networks to understand complex multi-layer regulation within biological systems. Interactive online databases are described that allow users to explore omics data, expand networks, and investigate relationships between biological entities and diseases.

The STRING database and related tools

The document discusses the STRING database and related tools for exploring protein-protein association networks, gene neighborhoods, phylogenetic profiles, and other computational predictions and experimental data. It notes that individual databases cover different species and formats, and have variable quality. STRING aims to integrate these resources using common identifiers, quality scores, and text mining while calibrating scores against experimental data and curated knowledge. Resources discussed include STRING for protein networks, STITCH for chemical networks, and COMPARTMENTS and TISSUES for subcellular localization and tissue expression data.

Systems biology - Bioinformatics on complete biological systems

This document discusses systems biology and bioinformatics. It describes how systems biology takes a holistic approach to study complete biological systems and all of their components and interactions. In contrast, earlier approaches in biology focused on studying one gene or protein at a time. The document outlines several key subfields and approaches within systems biology, including mathematical modeling of biological networks and pathways, data integration from various sources, and the use of association networks to predict functional relationships between biomolecules. It provides examples of publicly available databases like STRING and STITCH that compile interaction and association data from multiple sources for large numbers of organisms. The challenges of data integration are also discussed due to issues like incompatible identifiers and variable data quality across sources. The document then focuses on

STRING: Protein association networks

The document discusses protein association networks and the STRING database. STRING collects information on known and predicted protein-protein interactions for more than 9.6 million proteins, drawing from various sources including curated databases, experiments, text mining of literature, and gene neighborhood, gene fusion and co-occurrence data. It integrates this heterogeneous data into an association network with scored connections between proteins. The document instructs the user to explore the STRING database by searching for human insulin receptor and examining the different types of evidence and predicted associations.

STRING: protein association networks

The document discusses protein association networks and the STRING database. STRING integrates known and predicted protein-protein associations from various sources for over 9.6 million proteins, providing both a global view of protein networks as well as specific examples like the human insulin receptor. It summarizes the types of evidence that STRING uses, including curated databases, experiments, textmining, and inferred associations from genomic context and co-expression. The document also notes challenges in integrating data from different sources and databases.

Network biology: Large-scale integration of data and text

This document discusses natural language processing (NLP) techniques for extracting information from biomedical literature and integrating it with network and interaction data. It describes how NLP is used to identify entities like genes and proteins, extract relationships between entities, and integrate this text-mined information with existing interaction networks from databases like STRING to expand knowledge of protein interactions, complexes, pathways and associations with diseases. The document provides examples of using NLP analysis on sentences and the STRING and Tissues databases to explore tissue specificity and disease relationships for insulin and the insulin receptor.

Large-scale data and text mining - Linking proteins, chemicals, and side effects

This document discusses using data mining and text mining techniques to link proteins, chemicals, and side effects in molecular interaction networks. It provides examples of using the STRING and STITCH databases to explore protein and chemical networks. It also discusses how text mining of biomedical literature and electronic health records can help identify molecular interactions, adverse drug reactions, and support drug repurposing efforts.

Network biology - Large-scale integration of data and text

The document discusses network biology and integration of large-scale data and text to build interaction networks. It introduces the STRING database, which contains over 9.6 million proteins and integrates interaction data from curated databases, experiments, textmining, and predictive methods. The document uses human insulin receptor (INSR) as an example to demonstrate searching and analyzing the STRING network, showing evidence from different data sources for its interaction with IRS1. It also introduces other integrated networks in the STRING group including STITCH, COMPARTMENTS, TISSUES and DISEASES.

STRING & related databases: Large-scale integration of heterogeneous data

The document discusses the STRING database, which integrates heterogeneous biological data to generate association networks for proteins. It describes how STRING collects and connects curated knowledge, experimental data, and predicted interactions from genomic context, co-expression and text mining. The document also outlines exercises for users to explore protein-protein associations in STRING and related databases that integrate data on subcellular localization, tissue expression, and disease associations.

Large-scale integration of data and text

This document discusses large-scale integration of biological data and text mining. It describes three main parts: association networks that connect entities based on "guilt by association", protein interaction networks built using data from STRING and 2000+ genomes, and using genomic context like gene fusion, gene neighborhood, and phylogenetic profiles. It then provides examples of using STRING to query protein networks and discusses challenges of text mining like the exponential growth of literature and limitations of current natural language processing. Finally, it describes the Jensen Lab's approach of integrating curated knowledge, experimental data, predictions, and data from databases like STRING, STITCH, PubChem, COMPARTMENTS, Gene Ontology, UniProtKB, and disease databases into a common framework with

Network biology: Large-scale integration of data and text

The document discusses network biology and large-scale data integration. It describes protein-protein interaction networks like STRING that integrate data from curated knowledge, experiments, and predictions. It provides exercises to explore the human insulin receptor (INSR) in STRING, examining the types of evidence that support its interaction with IRS1. It also introduces other integrated networks like STITCH for chemicals and COMPARTMENTS for subcellular localization. Natural language processing techniques like named entity recognition, information extraction, and semantic tagging are used to integrate text data from the literature into these interaction networks.

Advanced bioinformaticsof proteomics datasets

This document discusses advanced bioinformatics approaches for analyzing proteomics datasets, including using signaling networks, association networks, and text mining. It describes using machine learning to predict protein interactions and developing scoring schemes to integrate data from multiple sources. The document also covers using text mining approaches like named entity recognition and information extraction to analyze the large amount of proteomics information available in scientific literature.

Text mining

The document discusses text mining and natural language processing techniques for extracting information from scientific documents, including information retrieval, entity recognition, and information extraction. It provides examples of identifying entities like genes and proteins, and extracting relationships between entities, such as regulatory interactions, from text. The techniques discussed aim to formalize facts and biological knowledge from unstructured text into a structured database for further analysis.

Biomedical literature mining

The document discusses various methodologies for extracting information from biological literature, including entity recognition to identify genes/proteins mentioned in text, relationship extraction using co-occurrence and natural language processing techniques, and text categorization to identify specific relationship types. It provides examples of applying these methods to extract entity and relationship information from a sample sentence.

Similar to Applied text mining (20)

Computational approaches to cell cycle analysis: Current research topics (tho...

Integration of biomedical literature and databases

Open access - making the most of biomedical literature mining

STRING: Large-scale data and text mining

Text mining for protein and small molecule relations

Network biology: Large-scale data and text mining

Systems biology - Understanding biology at the systems level

The STRING database and related tools

Systems biology - Bioinformatics on complete biological systems

STRING: Protein association networks

STRING: protein association networks

Network biology: Large-scale integration of data and text

Large-scale data and text mining - Linking proteins, chemicals, and side effects

Network biology - Large-scale integration of data and text

STRING & related databases: Large-scale integration of heterogeneous data

Large-scale integration of data and text

Network biology: Large-scale integration of data and text

Advanced bioinformaticsof proteomics datasets

Text mining

Biomedical literature mining

More from Lars Juhl Jensen

One tagger, many uses: Illustrating the power of dictionary-based named entit...

This document summarizes a Twitter thread discussing the uses of a dictionary-based named entity recognition tool called Tagger. Tagger can recognize genes, proteins, diseases and other biomedical entities. It is open source, runs quickly processing over 1000 abstracts per second, and achieves 70-80% recall and 80-90% precision. Tagger has been applied to tasks like identifying drug-disease associations, adverse drug events, and protein-protein interactions. It is available as a Docker container or web service.

One tagger, many uses: Simple text-mining strategies for biomedicine

The document summarizes a text mining tool called a tagger that can be used for named entity recognition in biomedical texts. It recognizes genes, proteins, chemicals, diseases, and other entities. The tagger is open source, runs quickly at over 1000 abstracts per second, and has 70-80% recall and 80-90% precision. It comes with Python and Docker implementations and can be accessed via a web service. It is useful for tasks like extracting functional associations from literature and electronic health records.

Extract 2.0: Text-mining-assisted interactive annotation

This document describes Extract 2.0, a text-mining tool that can assist with interactive annotation of documents. It uses dictionary-based tagging to identify relevant entities like genes and diseases. It achieves 70-80% recall and 80-90% precision on entity extraction and was evaluated in BioCreative challenges where it received positive feedback from curators. The tool is open source and available as a web service or Python wrapper.

Network visualization: A crash course on using Cytoscape

STRING & STITCH: Network integration of heterogeneous data

The document discusses STRING and STITCH, two online databases that integrate data on protein-protein interactions, pathways, and functional associations from various sources. STRING collects data on over 9.6 million proteins and 430 thousand chemicals from sources like text mining, experimental assays, and co-expression analyses. It aims to provide a comprehensive global view of known and predicted protein associations. STITCH also integrates interaction data but focuses more on chemical-protein interactions. Both databases provide user-friendly web interfaces for browsing and visualizing interaction networks.

Biomedical text mining: Automatic processing of unstructured text

1) Lars Juhl Jensen discusses biomedical text mining and automatic processing of unstructured text such as patent literature, grant proposals, FDA product labels, and electronic medical records. 2) Named entity recognition is used to identify genes/proteins, chemical compounds, diseases, and other entities in text through comprehensive dictionaries and flexible matching rules that account for variations. 3) Relation extraction uses natural language processing techniques like part-of-speech tagging and sentence parsing along with manually crafted rules and machine learning to identify implicit relations between entities in text such as transcription factor targets, kinase substrates, and protein-protein interactions.

Medical network analysis: Linking diseases and genes through data and text mi...

The document summarizes the work of Lars Juhl Jensen and others on medical network analysis and linking diseases and genes through data and text mining of electronic health records. It discusses how they have used Danish national health registries containing data on over 6 million patients and 119 million diagnoses over 14 years to study disease trajectories and comorbidities. It also describes how they have developed methods to integrate data from various sources to generate networks linking diseases and genes.

Network Biology: A crash course on STRING and Cytoscape

This document provides an overview of STRING, a protein-protein association database, and Cytoscape, a network visualization tool. It describes how STRING contains functional associations between proteins derived from genomic context, co-expression and curated databases. Cytoscape can import STRING networks and external data to map onto nodes. It offers visualization of networks through layouts and attributes, and analysis through clustering, selection filters and enrichment. The document recommends using these tools together to explore protein association networks.

Cellular networks

This document discusses different approaches to visualizing cellular networks and the molecular interactions between proteins. It notes that there are many different types of data that could be shown, such as protein names, functions, localization, expression, modifications, and interaction types. However, it is impossible to show all this information at once. The document recommends using different visualizations like force-directed layouts to distribute proteins in 2D or lining up interactions in 1D. It acknowledges open challenges like showing time-course data and modification sites. In the end, the document thanks several researchers who have contributed to mapping and visualizing cellular networks.

Cellular Network Biology: Large-scale integration of data and text

The document discusses various community resources and software tools for integrating large-scale data and text, including STRING for protein networks, STITCH for chemical networks, COMPARTMENTS for subcellular localization, TISSUES for tissue expression, and DISEASES for disease associations. It provides an overview of text mining techniques used to extract information from literature to build networks in these resources. The presenter demonstrates the Cytoscape App which can import and analyze networks from STRING, perform queries, and analyze subcellular localization, tissue expression, and disease enrichment.

Statistics on big biomedical data: Methods and pitfalls when analyzing high-t...

This document discusses statistical methods for analyzing high-throughput biomedical screens and common pitfalls. It introduces several statistical tests such as t-tests, ANOVA, Fisher's exact test, and the Mann-Whitney U test. It also discusses challenges like multiple testing, resampling techniques, and biases that can occur like studiedness bias and abundance bias in big data analyses. Controlling false discovery rates and considering effect sizes are recommended over solely relying on p-values to determine biological significance.

Tagger: Rapid dictionary-based named entity recognition

Tagger is a named entity recognition tool that can process over 1000 abstracts per second using a dictionary-based approach. It achieves 70-80% recall and 80-90% precision using comprehensive dictionaries, expansion rules, and a curated blacklist to identify entity types like genes, proteins, chemicals, and diseases. The tool has a C++ engine, is inherently thread-safe, and includes interactive annotation, Python wrappers, and a REST API.

Network Biology: Large-scale integration of data and text

Lars Juhl Jensen leads a group that conducts large-scale integration of biological and medical data using proteomics, text mining, and medical data mining. The group develops protein interaction networks, disease networks, and association networks. They collaborate internationally on projects involving over 9.6 million proteins and 2000 genomes. The group works to integrate data from many sources in different formats to build comprehensive networks and knowledgebases, and also mines biomedical text to link genes and proteins with diseases.

Medical text mining: Linking diseases, drugs, and adverse reactions

This document discusses medical text mining and linking diseases, drugs, and adverse reactions. It describes using text mining on clinical narratives in Danish to recognize named entities like drugs and diseases, identify relationships between them like adverse drug reactions, and discover new ADRs. The goal is to generate structured data on topics like comorbidities, diagnosis trajectories, and reimbursement to supplement limited structured data and help busy doctors by analyzing large amounts of unstructured text.

Medical data and text mining: Linking diseases, drugs, and adverse reactions

This document discusses medical data and text mining to link diseases, drugs, and adverse reactions. It describes using structured data from Danish central registries and unstructured data from hospital electronic health records. Named entity recognition is used to extract diseases, drugs, and adverse reactions from free text clinical notes written in Danish. Hand-crafted rules are developed to identify relationships between extracted entities like adverse drug reactions. This allows estimating frequencies of known adverse drug reactions and discovering new adverse drug reactions by analyzing diagnosis trajectories and medication information.

Cellular Network Biology

This document discusses cellular network biology and summarizes several key papers on topics like proteome analysis using mass spectrometry, integrating protein network and experimental data, challenges with different biological databases having varying formats and quality, and using natural language processing techniques like named entity recognition and relation extraction to analyze medical text for information like diagnosis trajectories and adverse drug reactions.

Biomarker bioinformatics: Network-based candidate prioritization

The document discusses three parts of biomarker bioinformatics: data integration from multiple databases, text mining of scientific literature, and using that integrated data to prioritize biomarker candidates. It describes combining data on 9.6 million proteins from curated databases, using text mining to extract named entities from over 10,000 papers, and then using network and heat diffusion approaches to rank candidates based on evidence in the integrated data. The goal is to help identify new biomarker candidates from large amounts of biological data.

The Art of Counting: Scoring and ranking co-occurrences in literature

The document discusses methods for scoring and ranking co-occurrences of entities like diseases and genes in literature. It describes counting co-occurrences within different text levels like documents, paragraphs and sentences, and using techniques like z-score transformations and weighted combinations that can rank entities for a given query without changing the overall ranking. The methods have been implemented in web tools that can return results for queries within seconds using preprocessed named entity recognition results stored in a relational database.

Text-mining-based retrieval of protein networks

This document describes a method for using text mining of biomedical literature to retrieve protein networks. Key aspects include using text mining and named entity recognition on sets of abstracts from PubMed queries to identify proteins of interest and their relationships, then constructing a protein interaction network. This network can then be explored and visualized using the Cytoscape App integration of the text mining approach within the STRING database framework.

Medical data and text mining: Linking diseases, drugs, and adverse reactions