GWeb

GWeb is a hybrid Web application to run jobs for genomic analysis using Genestrip.

License

GWeb is free for non-commercial use. Please contact daniel.pfeifer@progotec.de if you are interested in a commercial license.

Building and installing

GWeb is structured as a standard project for Maven 3.5 or higher with modules and is compatible with the JDK 1.8 or higher. However, generating the local variant of GWeb requires the JDK 11 or higher on a Windows machine as it uses JLink.

GWeb is implemented as a hybrid Web application such that

• it can be run on a Servlet-based Web server (tested under Tomcat 10),
• it can be installed and executed entirely locally on a Windows machine.

Web server variant

mvn clean install will compile and test GWeb and generate the war file ./gweb-ui/target/gweb.war, which may be deployed under Tomcat for example. The deployment requires a Postgres database (tested under Postgres 14) to be in reach under the JDBC URL jdbc:postgresql://localhost/gweb with a Postgres admin user postgres and password postgres. You may configure this via the file ./gweb-ui/src/main/webapp/META-INF/context.xml.

Local variant

To build the local variant of GWeb, you must set the property javapackager.jdkPath in ./gweb-launcher/pom.xml to the path of your corresponding JDK 11 (or higher) on a Windows system. mvn -P winexe clean install will then (eventually) create the file ./gweb-launcher/target/gweb-??-windows-x64.zip which contains the entire local application (including a stripped but sufficient JRE).

To install gweb-??-windows-x64.zip just unzip it and the folder gweb will appear. Start the file gweb/bin/GWeb.exe via double click or from the command line from within the folder gweb/bin. (You may place the folder gweb with its contents anywhere you want for this.) GWeb.exe starts a local Web server on port 81 and opens the URL http://localhost:81 in your standard Web browser. The UI of GWeb will appear with the admin user logged in by default.

The local installation uses a local, file-based HSQL database, which stores its data under gweb/data. Any other application data, such as genomic databases and analysis files are also kept under gweb/data. To migrate from one version of GWeb to the next, you may just backup and copy the folder data to a new installation folder gweb. When started, the old application data should all be present again in the new installation.

A ready-made zip file of the local variant is available under https://www.genestrip.org/files/bin/.

Functionality

GWeb's main functionality is to run jobs for genomic analysis using Genestrip.

The main features are:

• creating and executing jobs to analyze fastq files,
• creating URLs and paths for downloading fastq files during analysis or placing them in local file system,
• installing related genomic Genestrip databases for use during analysis,
• creating users entries and assign roles to them to handle user rights,
• creating persons entries associated with users,
• viewing and downloading result files from analysis,
• viewing and downloading information files regarding Genestrip database contents.

User roles

There are four user roles, where stronger roles subsume the rights of a weaker roles:

• "No Login" - the user must not login. This is mainly to deny a user access but keeping his application data in the system,
• "Viewer" - may view job results assigned to her / him and view Genestrip database information,
• "Job Executor" - may create and execute jobs and view related results but only based on her / his application data,
• "Admin" - may access and do everything, including user and person management.

Default user

The system configuration allows for defining a default user who is logged in be default. Regarding the local variant of GWeb, the (initial) default user admin with password admin has the role "Admin". You may always logout the default user, which allows you to login as a different user.

Supplying fastq files and Genestrip databases

There are three ways to supply a job with fastq files:

• Via a URL added in the tab "FastQ URLs": The URL must point to a respective fastq file (with public access). The file will be streamed for analysis and will not be stored by GWeb.
• Via a local folder on the server-side: This option is particularly useful under the local variant of GWeb. The location of the folder is user-specific and can be displayed with a respective UI button in the job details. Once a fastq file is placed inside the folder, GWeb will detect it and offer it in the UI for selection in the job details.
• When configured accordingly, users under the roles "Admin" or "Job Executor" may set server-side paths to fastq files in the "Resource" tab. This option supports glob patterns in paths described here. All fastq files matching the path pattern will be jointly analyzed. Obviously, this requires fastq files to present in respective locations on the server-side. This option is particularly useful under the local variant of GWeb.

Two entries for fastq files may be added per job at max. All matching fastq files will be jointly analyzed during that job.

GWeb does not offer an upload of fastq files from the client-side (yet) - mainly to avoid high disk storage consumption on the server.

Technical documentation

Basic architecture

GWeb is light weight and uses no frameworks. On the server-side, it is based on the following Java-related technologies:

• Servlets and JSP (based on the Jakarta Servlet API),
• JDBC (using SQL-statements compatible with Postgres and HSQL),
• JAX-RS (using the Jersey reference implementation as a pure library),
• Genestrip with its dependent libraries.

GWeb (re-)uses worker threads for executing Jobs on the server-side. It queues jobs using a Timer such that only one job is executed at a time. To do so, Genestrip is used as a library and invoked in-process. (This implies that your Web server may require a lot of heap but also CPU time during analysis.) The number of worker threads is configurable.

Heap consumption is mainly determined by the size of the Genestrip database, as the database must be loaded into memory for a corresponding analysis job. To save heap space, only one Genestrip database (the one for the current job) is kept in memory at a time.

On the client-side, GWeb is based on HTML 5, CSS and JavaScript without additional JavaScript libraries. The GUI is a thin client without page switches. Data exchange with the server works entirely via REST. The GUI was tested for compatibility on several modern Web browsers.

Location of application data

Application data including

• installed Genestrip databases,
• analysis files from jobs,
• log files from jobs and
• database information files

are stored in the folder data (or in respective sub-folders). The path to data must be configured. For the local variant of GWeb the data folder is in the installation folder of gweb.

All other application data is stored in a relational database (using Postgres or HSQL). For the local variant of GWeb the HSQL database files are also under the folder the data.

Configuration

The configuration of the Web server variant must all be done in ./gweb-ui/src/main/webapp/WEB-INF/web.xml via init-param settings (for the Servlet ...GenestripRestApplication) and one context-param setting. The shipped web.xml shows where to place these settings accordingly.

The local variant of GWeb can be configured in ./gweb-local/src/main/webapp/WEB-INF/web.xml but that is rarely necessary.

Obviously, rebuilding and reinstalling is required afterwards.

The following table describes the settings.

Name	Category	Type	Default Value	Description
localInstall	init-param	boolean	false	Whether a user not in an "Admin" role may see the server-side path to their fastq file folder. This should only be allowed set to true for local variant of GWeb
defaultUser	init-param	String	null	Login name of the default user that is automatically logged in when accessing the GWeb UI. The user must have a database entry for this to work. If not set, the default user functionality is off.
initDefaultUser	init-param	boolean	true	Whether the default user should automatically be created with the role "Job Viewer" during Servlet initialization. If the default user already exists, it will neither be created nor changed.
jobDelay	init-param	long	1000	Time in ms to check the job queue for the execution of the next job (when idle).
jobPeriod	init-param	long	10000	Waiting time in ms for the job queue to execute the next job after the former job was finished.
initAdmin	init-param	boolean	true	Whether the user admin with password admin and role "Admin" should automatically be created during Servlet initialization. If the user admin already exists, it will neither be created nor changed.
genestripBaseDir	init-param	String	./data	The server-side path to the folder data. It must exist upfront and the Web server must have write access to it.
sqlDialect	init-param	String	POSTGRES	Only HSQL and POSTGRES are valid values and must match the connected database system type.
threads	init-param	integer	-1	The number of consumer threads of Genestrip used during analysis of fastq files.
initDBs	init-param	boolean	true	Whether to create a set of URL entries for Genestrip databases as given by Genestrip DB. The entries will be added as part of the initial database schema creation only. (Installation must still be done manually by clicking a respective UI button.)
filePathRole	context-param	String	null	The minimum role in which a user can set server-side paths to fastq files in the UI's "Resource" tab. Possible values are ADMIN and RUN_JOBS. If not set, the corresponding feature is disabled.