Biomedical Annotation by Humans and computers in a Keyword-driven world
From the website:
Abstract:
As part of our project with the NCBO we have been curating expression experiments housed in NCBI’s GEO data base and annotating a variety of rat-related records using the NCBO Annotator and more recently, mining data from the NCBO Resource Index. The annotation pipelines and curation tools that we have built have demonstrated some strengths and shortfalls of automated ontology annotation. Similarly our manual curation of these records highlights areas where human involvement could be improved to better address the fact that we are living in the Google era where findability is King.
Speaker Bio:
Simon Twigger currently splits his time between being an Assistant Professor in the Human and Molecular Genetics Center at the Medical College of Wisconsin in Milwaukee and exploring the iPhone and iPad as mobile platforms for education and interaction. At MCW he has been an investigator on the Rat Genome Database project for the past 10 years, he worked with the Gene Ontology project and has been active in the BioCuration community as co-organizer of the past three International BioCuration meetings. He is the former Director of Bioinformatics for the MCW Proteomics Center and was previously the Biomedical Informatics Key Function Director for the MCW Clinical & Translational Science Institute. He is a Semantic web enthusiast and is eagerly awaiting the rapture of Web 3.0 when all the data will be taken up into the Linked Data cloud and its true potential realized.
Annotation, useful annotation anyway, is based on recognition of the subject of annotation. Should prove to be an interesting presentation.
Notes from the webinar:
(My personal notes while viewing the webinar in real time. The webinar controls in all cases of conflict. Posted to interest others in viewing the stored version of the webinar.)
Rat Genome Database: http://rgd.mcw.edu / interesting questions that researchers ask / Where to find answers, PubMed 20 million+ citations, almost 1 per minute / search is the critical thing – in all interfaces / “Being able to find information is of great importance to researchers.” / NCBO Annotator www.bioontology.org/wiki/index.php/Annotator_Web_service / records annotated – curated the raw annotations – manual effort needed to track it down – / rat strain synonyms has issues / work flow description / mouse gut maps to course (ex. of mapping issue) / Linking annotations to data / RatMine faceted-search + lucene text indexing , interesting widgets / – Driving “Biological” Problem Part 2 – 55.6 % of researchers rarely use archival databases, 56.0% rarely use published literature / 3rd International biocurator meeting Amos Bairoch – “trying to second guess what the authors really did and found.” / post-publication effort to make content be found. different from academic model where publication simply floats along. / illustration of where the annotation path fails and the consequences of that failure. / very cool visualization of how annotations can be visualized and the value thereof / put in keywords and don’t care about it being found (paper) , NCBO Resource Index could be a “semantic warehouse” of connections, websites: gminer.mcw.edu, github.com/mcwbbc/, bioportal.bioontology.org, simont -at- mcw.edu @simon_t