Skip to content

based on annotator.js, an annotation framework enable user account and annotation permission management and templating annotation plugin in biomedical domain.

License

Notifications You must be signed in to change notification settings

dbmi-pitt/dbmi-annotator

Repository files navigation

AnnotationPress

Overview

AnnotationPress is designed to provide a user friendly tool that helps scientists annotate scientific claims, data, and evidence in HTML and PDF documents. The current version is configured to annotated drug-drug interactions in both full text journal articles and structured product labeling. This is a particularly challenging domain that required the development of several annotation workflows which we think will be useful for for other scientific domains. A key aspect of AnnotationPress is its use of annotation plug-ins which make the tool is easily adaptable to other annotation use cases. Another distinctive feature is that AnnotationPress implements the Micropublication and Open Annotation data standards. These standards specify how to relate annotations in target documents about data, methods, and materials to scientific claims.

If you are new to AnnotationPress, please watch our video that show two annotation plugins in action.

AnnotationPress is an Apache licensed open source project available on github. Please contact us if you are interested in contributing to its development, if you would like to request new features, or have any other questions.

Licensing

AnnotationPress is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.

How to run - Production

  1. Install docker(v1.12.3), docker-compose (v1.8.1) and apache (v2.4) server

    Apache2 server configuration

    # AnnotationPress Docker configuration ###############################

    # proxy for docker exposed service at port 8085

    ProxyPass /dbmiannotator http://localhost:8085/dbmiannotator RewriteRule /dbmiannotator(.*) http://localhost:8085/dbmiannotator$1 [P]

    # proxy for docker exposed service at port 8085 for annotator store

    ProxyPass /annotatorstore http://localhost:8085/annotatorstore RewriteRule /annotatorstore(.*) http://localhost:8085/annotatorstore$1 [P]

    # stylesheet folder relocate at Document root(dailymed, PMC, wiley)

    RewriteRule /dbmiannotator/dailymed(.*) http://localhost/DDI-labels/dailymed$1 [P] RewriteRule /dbmiannotator/PMC(.*)_files/(.*) http://localhost/PMC/PMC$1_files/$2 [P] RewriteRule /dbmiannotator/wiley(.*)_files/(.*) http://localhost/wiley/wiley$1_files/$2 [P]

  2. Create local volume for postgres and elasticsearch data mount point:

    $ docker volume create --name postgres-volume -d local
    $ docker volume create --name elasticsearch-volume -d local
    $ docker volume create --name elastic-snapshot-volume -d local
    
  3. Run dbmi-annotator with all dependencies in docker container:

    $ git clone
    $ cd docker-dbmi-annotator/dbmi-annotator
    $ docker-compose up
    

Optional: pre-load annotations from csv to DB mpevidence

$ python loadDomeoAnnsToRDB.py <pghostname> <pgport> <pguser> <pgpassword> <clean existing data (1: yes, 0: no)>

ex. $ python loadDomeoAnnsToRDB.py postgres 5432 dbmiannotator dbmi2016 1

  1. optional: clean all data in table
  2. preprocess csv, generate preprocess-domeo.csv
  3. load data into schema mpevidence

Optional: pre-load annotations from postgres DB mpevidence to elasticsearch

$ docker exec -it dbmi-annotator bash $ cd translation/rdb-data-loader $ python load-rdb-annotations.py <pg hostname> <pg username> <pg password> <es hostname> <annotation author>

ex. python load-rdb-annotations.py postgres dbmiannotator dbmi2016 elasticsearch [email protected]

For how to build production mode via docker: https://github.com/dbmi-pitt/docker-dbmi-annotator/blob/master/README

How to run - Development

Dependences:

  1. Elasticsearch version 1.7 running on port 9200

  2. Annotator Store running on port 5000

  3. Postgres DB running on port 5432:

    create database dbmiannotator

    SQL script that create dbmiannotator schema: https://github.com/dbmi-pitt/dbmi-annotator/blob/master/db-schema/rdb-postgres-schema.sql

    SQL script that initialize plugin settings: https://github.com/dbmi-pitt/dbmi-annotator/blob/master/db-schema/rdb-postgres-initial.sql

  4. Apache2 server running on port 80

Installation:

1. nodejs packages $ npm install

2. compile browser side JS packages (rerun when made modifications on config/config.js) $ browserify app.js -o public/dbmiannotator/js/app.bundle.js

Configuration:

1.Create config.js $ cp config/config.sample.js config/config.js

2.Edit config.js based on system environment

3.Apache2 configurations:

#proxy for local nodejs server on port 3000
RewriteRule /dbmiannotator$ http://localhost:3000/dbmiannotator [P]
RewriteRule /dbmiannotator/(.*) http://localhost:3000/dbmiannotator/$1 [P]

#proxy for local annotator store on port 5000
ProxyPass /annotatorstore http://localhost:5000/
RewriteRule /annotatorstore(.*) http://localhost:5000$1 [P]

Run server:

$ cd dbmi-annotator
$ nodemon server.js (run '$ npm install -g nodemon', if command is not available)
$ service apache2 start

access AnnotationPress through ``'http://localhost/dbmiannotator'``

Add documents to annotate:

  1. Download the HTML using your browser to some place e.g., /path/
  2. If needed, rename the file to a more appropriate name. E.g., for PMC articles, it is good to concatenate 'PMC' with the article's PMC identifier. For example, for PMC 1459289 you would:

$ mv /path/file.html /path/PMC1459289.html

$ mv /path/file_files /path/PMC1459289_files

  1. Add html document and related files to apache2 home directory that is relevant for the papers publisher. E.g., for PMC:

$ cp -r /path/PMC1459289* /var/www/html/PMC/

  1. Register new document by opening up in a text editor the relevant file in the folder dbmi-annotator/article-list/ . For example, for PMC:

$ sublime article-list pmc-list.csv

<you local article identifier> http://localhost/PMC/PMC1459289.html

  1. Confirm that you can access the article in your browser using the link e.g., http://localhost/PMC/PMC4536363.html
  2. Now, refresh the AnnotatorPress in your browser and you should see the new article under the relevant publisher. Click to open.

This should work for all of the publishers we have ensured compatibility between HTML versions of their papers (with their associated files) and our style sheets:

Dailymed PMC Wiley Elsevier Springer Sage Taylor & Francis Wolters Kluwer

Directories

Node.js based program - folder structures

config/ - configurations models/ - represents data, implements business logic and handles storage controllers/ - defines your app routes and their logic public/ - contains all static files like images, styles and javascript views/ - provides templates which are rendered and served by your routes server.js - initializes the app and glues everything together package.json - remembers all packages that your app depends on and their versions

reference: https://www.terlici.com/2014/08/25/best-practices-express-structure.html

design

(1) design/diagram-workspaces/ keep source code of diagram

pencil prototyping tool: *.ep dia: *.dia yEd: *.graphml

(2) design/images/ exported images for software design

About

based on annotator.js, an annotation framework enable user account and annotation permission management and templating annotation plugin in biomedical domain.

Resources

License

Stars

Watchers

Forks

Packages

No packages published