About

This project starts with rebranding the codebase of CTD^2 Dashboard project. It encapsulates three modules:

Core: contains the main data structures and basic DAO methods.
Admin: contains importers/normalizers/converters for populating the database (depends on core)
Web: provides a Web-based API for querying the database; also contains a thin-client (known as Signature Web UI) that helps with basic navigation

Install

The code is structured as a Maven project. We suggest you work in an environment with Java 11, MySQL 5.7.x and Maven 3.x.x properly installed. You will need two databases: one for the production, by default hipc, and one for the testing, again by default hipc_test.

Before running any mvn jobs, make sure you do the basic configuration. Example configuration files are provided within the distribution. The easiest way to do the configuration is as follows:

# this is for Log4j properties, no need to edit unless you want your logs to be saved
cp -f core/src/main/resources/log4j.properties.example core/src/main/resources/log4j.properties
# the following contains main properties (including database user/password)
# please make sure you edit minimally the USER/PASSWORD pair and the TEST/MAIN database URLs.
cp -f core/src/main/resources/META-INF/spring/signature.properties.example core/src/main/resources/META-INF/spring/signature.properties

After you are done with the configuration, you will be good to go ahead for a basic install:

mvn clean install

The command above will compile the project, run the tests and create the necessary jar/war files for you. Make sure the tests are sucessfully completed. From here on you can follow the module-specific notes below.

Notes about modules

Core: Data structures and DAO methods

This module tries to capture the main database schema outlined below:

We are taking advantage of factory pattern and also separating class definitions (interfaces) and their actual implementations (*Impls) for convenience. We basically try to convey the following basic ideas in our implementation:

Every item that is going to be persisted in the database is of type DashboardEntity.
Every DashboardEntity has a unique id, which is autogenerated by the database (i.e. when persisted in the database); therefore id spaces do not clash.
All new instances of the classes should be created via DashboardFactory -- this is for better handling of the semantics.

So let's take Compound as an example. The class Compound is actually an interface that extends the Subject interface which, in turn, extends the DashboardEntity interface. The interface determines which methods a class should provide. The actual implementation of the methods goes into the class CompoundImpl which, programmatically speaking, implements Compound. This is better explained with the following simple UML diagram:

The following UML diagram also shows properties and methods in a detailed manner:

Because of these implementation choices, the good practice for creating a new instance of a specific object is as follows:

// Demonstration of how a new instance of an object should be created

// Create a factory
DashboardFactory dashboardFactory = new DashboardFactory();

// Factory grabs the actual implementation and binds it to the object
Compound compound = dashboardFactory.create(Compound.class);
// Developer modifies the object through the methods defined in the interface
String pyrethrinII = "COC(=O)C(\\C)=C\\C1C(C)(C)[C@H]1C(=O)O[C@@H]2C(C)=C(C(=O)C2)CC=CC=C";
compound.setSmilesNotation(pyrethrinII);
compound.setDisplayName("Pyrethrin II");

Furthermore, the interfaces do not know about the persistence details -- that is, column/table names and etc. We do not empose a pre-set SQL schema onto the persistence layer, but instead let Hibernate deal with the details -- i.e. creating the schema. So all Hibernate and Persistence annotations go into the Impl classes which use the actual interfaces as Proxy (see the UML diagram above). The specific details -- database dialect, usernames, passwords, database names -- are all defined in the signature.properties file mentioned above.

Basic querying methods are implemented as part of the DashboardDao. Developers do not have to deal with the configuration and/or initialization of the Dao class, but rather they can get it through Spring facilities, e.g.:

ApplicationContext appContext = new ClassPathXmlApplicationContext("classpath*:META-INF/spring/applicationContext.xml");
DashboardDao dashboardDao = (DashboardDao) appContext.getBean("dashboardDao");

DashboardDao can be used to persist objects:

Synonym synonym = dashboardFactory.create(Synonym.class);
synonym.setDisplayName("Synonym 1");
dashboardDao.save(synonym);

Normally, you don't have to worry about the id field. When an object is created via DashboardFactory, it has an id of null. When it gets persisted, the id field is updated with the automatically-generated value:

Synonym synonym = dashboardFactory.create(Synonym.class);
assert synonym.getId() == null;
dashboardDao.save(synonym);
assert synonym.getId() != null;

DashboardDao can also be used to acquire objects that are already in the database:

// Grabs all compounds in the database
List<Compound> allCompounds = dashboardDao.findEntities(Compound.class);

// or more generic entities, for example all Subjects
List<Subject> allSubjects = dashboardDao.findEntities(Subject.class);

or to conduct a find operation:

List<Compound> compoundsBySmiles = dashboardDao.findCompoundsBySmilesNotation(pyrethrinII);

Admin: Importers/Converters and major database operations

This module provides a command-line interface (CLI) for users to populate the database via utilization of data importers. When mvn install is called either on this module or on the parent module, a JAR file with all its dependencies included is generated under the default target/ directory. This JAR file can be run via the following command:

java -jar admin/target/dashboard-admin.jar

For example:

java -jar admin/target/dashboard-admin.jar -h

will list all available options to you. For more information about running the admin tool, please visit the Signatures Admin Tool page.

Web: HIPC Signature UI

Deploy

The web application is now built as a spring boot application. To run it, use the follow command:

java -jar web/target/hipc-signature-spring-boot.war

and point your browser to http://localhost:8080.

You can also launch it by mvn -f web/pom.xml spring-boot:run during development and testing.

In case we need to deploy to a tomcat with a regular war file, we should roll back the change mainly of commit 73fe78e.

Background data

Gene

When we release a version of data, the background data of genes should be updated first from ftp://ftp.ncbi.nih.gov/gene/DATA/GENE_INFO/Mammalia/

Two files are used:

Homo_sapiens.gene_info.gz
Mus_musculus.gene_info.gz

Cell Subset

Data source: https://raw.githubusercontent.com/obophenotype/cell-ontology/master/cl.owl

License

This project is licensed under the BSD 3-Clause License - see the LICENSE file for details

floratos-lab/hipc-signature