Skip to content

NoB Use Cases

Jump to bottom Edit New page
goodb edited this page Sep 22, 2014 · 39 revisions

This is an archive of the NoB Use Cases submitted for the first hackathon (Stanford April, 2014). Please see the new Google doc for ideas proposed for the next hackathon (San Diego, November, 2014)

For the moment, do NOT edit below...

Identification and visualization of subnetworks of genes that are expressed and co-regulated

  • description: post
  • proposer: Melissa Haendel (in absentia)

Extraction gene-drug-disease-etc. relationships from clinicaltrials.org

  • description: post
  • proposer: Obi Griffith (in absentia)

N of 1 cancer description

  • description: post
  • proposer: Ben Ainscough

Variant annotations

  • description: post
  • proposer: Chunlei Wu

Resource Identification Initiative

Collaborative Ontology Term Mapping resource

  • description: While many term-to-term mapping resources do exist, many more are needed, automated techniques are not perfect, and manual curation remains the gold standard. We propose a web application that would enable manual ontology term mapping as well as capturing the created mappings for re-use. It would build on existing resources such as Monarch and NCBO BioPortal.
  • proposer: Ben Good, Michel Dumontier, Andrew Su.

A crowdsourced triple rating service

  • description: Input: triples mined from text, Output: triples rated for believability, evidence, interestingness. candidate implementations: mechanical turk, game, web/mobile.
  • proposer: Ben Good

Text mining drug-disease association verification

  • description: Given a prediction of an association between a drug and a disease (so either a drug-indication association, or drug-adverse event association), examine the existing public data, to give a thumbs up or thumbs down.
  • proposer: Nigam Shah (in absentia)

Network-based verfication of predicted drug-drug-event triples

  • description: Given a predicted list of drug-drug-event triples, assign a plausibility score by analyzing the network of biothings. ~5900 predictions at: http://goo.gl/kA2Uph
  • proposer: Nigam Shah (in absentia)

An interface for browsing networks of biothings with a dial for controlling the confidence (e.g. precision/recall) of the presented edges.

  • description: Input: triples with confidence ratings Output: human interface for browsing (start with simple text search) candidate implementations: web
  • proposer: Ben Good

Stable, fast, accurate text mining Web service for extracting gene-gene interactions from text.

Network-based identification of medical devices linked to adverse events

  • description: Analyze the NoB and make a list of medical devices that are associated with adverse events, and sort them for further review
  • proposer: Nigam Shah (in absentia)

Click-graph analysis of Wikipedia to identify synonyms

  • description: Analyze the click-graph of wikipedia accesses, and learn common synonyms for medical concepts. (e.g. what searches land you on the page for myocardial infarction).
  • proposer: Nigam Shah (in absentia)

Finding drugs for target proteins

  • description: Interrogate the NoB for drugs that could modulate a set of targets. Identify key features about these drugs -> chemical functional groups, known indications, known adverse events, and pathways.
  • proposer: Michel Dumontier

Mapping EHR lab codes to molecular biothings

  • description: Given a set of lab prescription code labels (such as Glucose, Urine) coming from EHR, define good quality mappings to an ontology (LOINC?), then to relevant LOD.
  • proposer: Adrien Coulet, Michel Dumontier

Concept recognition in text using patterns combining syntactic and semantic elements

  • description: Identify biomedical ontology concepts in text. The CRAFT corpus contains gold standard manually created semantic annotations of concepts from 8 biomedical ontologies (CL, ChEBI, GO:BP/MF/CC, SO, PRO, NCBITaxon) as well as syntactic annotation (sentence segmentation, tokenization, POS, PennTreeBank) and can be used for training and evaluation. Sub-projects include:
    • Defining a pattern language for combining syntactic and semantic pattern elements
    • A tool for visualizing a set of patterns matching (or not) against positive and negative examples (like a regex tester)
    • Learning patterns from training data
    • Evaluate different measures of precision and recall (for example, semantic similarity vs. all-or-nothing evaluation) for effects on training and evaluation.
  • proposer: Mike Bada, Kevin Livingston

Pathway Interaction Palette

  • description: Identify basic interaction types needed for Wikipathways to describe biological pathways. We started defining a new palette for drawing biological interactions in PathVisio/Wikipathways by choosing basic interaction types and mapping them to interactions in other standard pathway description languages.
  • progress : https://drive.google.com/?tab=mo&authuser=0#folders/0BzT8IiFDIm5DRDRPWC1UeG1Zdm8
  • proposer : Anwesha Bohler and Martina Kutmon.

Link diseases and pathways

We use disgenet, wikipathways(human) and bridge db. To determine for each pathway the enriched diseases and other way round.

Nanoparticle corona proteins to biothings

###Integration of genetic variation in gene-disease interaction

  • What we need: Integration of genetic variation in gene-disease interaction (see DisGeNET http://ibi.imim.es/web/DisGeNET/v01/). The possibility to analyse a gene-disease network (like you can do with DiGeNET, using Cytoscape) could be improved with the integration of genetic variation information. This information could be integrated in the network analysis using two different approaches: Bottom-up and Top-down. Approach Bottom-up : Genetic variations-gene-pathway-disease Starting with the GENETIC VARIATION we need an unifying resource that compiles/integrates:
  1. How a variant can impact on the gene function (Synonymous, non synonymous…)
  2. How this change can impact the gene-gene interaction inside pathway.
    It could be useful translate the previous points in a gene score that should be incorporated in the gene-disease network analysis (like DisGeNET). Approach top-down: Disease-pathway- gene –genetic variations The analysis starts from the DISEASE –GENE analysis network and the genes involved in a specific disease should be analysed inside their pathways . At this point the approach should include also the genetic information affecting genes and their pathway. This two approaches should be combined together to get a better idea of the biological meaning of the elements involving in the disease: genetic variations, genes and pathways.
  • How use cases could help our needs. 1.Identification and visualization of subnetworks of genes that are expressed and co-regulated.
    Useful to increase the integration inside the disease-gene network analysis. 2.Variation annotation: Useful tool that give a referent unified source of the genetic variation environment, it could permit to not miss information. It should have the functional impact, clinical annotations, the pubmed reporting the variant, epidemiological data. 3.Stable, fast, accurate text mining Web service for extracting gene-gene interactions from text. Useful for the step of the prediction of gene-gene interaction after considering the genetic variation influence. We would like to suggest also text-mining services that allow to further characterize gene-disease associations.

PubMed++

  • Proposers: Jonathan Mortensen
  • Goal: Augment pubmed search results with NoB data
  • This use-case provides a nice, constrained project for the Hackathon
  • Components:

Framework and formal model of multiscale phenotype

  • Proposers: Ivo Georgiev
  • Goal: Create a hierarchical framework for phenotype description on all physiological scales
  • Too big for a hackathon use case itself but many use cases can be extracted from it
  • Proposal page - comments are most welcome!
Add a custom footer
Add a custom sidebar
Clone this wiki locally