blend4j is a partial reimplemenation of the Python library bioblend for the JVM. bioblend for Python is a library for scripting interactions with Galaxy, CloudMan, and BioCloudCentral.
For simplicity, these examples assume the following import statements are used to import all of blend4j's Galaxy functionality.
import com.github.jmchilton.blend4j.galaxy.beans.*;
import com.github.jmchilton.blend4j.galaxy.*;
Listing current users history:
GalaxyInstance galaxyInstance = GalaxyInstanceFactory.get(url, apiKey);
HistoriesClient historiesClient = galaxyInstance.getHistoriesClient();
for(History history : historiesClient.getHistories()) {
String name = history.getName();
String id = history.getId();
String message = String.format("Found history with name %s and id %s", name, id);
System.out.println(message);
}Find a data library by name and print its contents:
final GalaxyInstance galaxyInstance = GalaxyInstanceFactory.get(url, apiKey);
final LibrariesClient librariesClient = galaxyInstance.getLibrariesClient();
final List<Library> libraries = librariesClient.getLibraries();
Library testLibrary = null;
for(final Library library : libraries) {
if(library.getName().equals("test-library")) {
testLibrary = library;
}
}
if(testLibrary == null) {
return;
}
for(final LibraryContent content : librariesClient.getLibraryContents(testLibrary.getId())) {
final String type = content.getType(); // file or folder
final String name = content.getName();
final String id = content.getId();
final String message = String.format("Found library content of type %s with name %s and id %s", type, name, id);
System.out.println(message);
}The following code demonstrates running a workflow. It assumes the existence of a workflow named TestWorkflow1 with two inputs labeled WorkflowInput1 and WorkflowInput2. Additionally it assumes a history named TestHistory1 already exists containing two datasets Input1 and Input2.
final GalaxyInstance instance = GalaxyInstanceFactory.get(url, apiKey);
final WorkflowsClient workflowsClient = instance.getWorkflowsClient();
// Find history
final HistoriesClient historyClient = instance.getHistoriesClient();
History matchingHistory = null;
for(final History history : historyClient.getHistories()) {
if(history.getName().equals("TestHistory1")) {
matchingHistory = history;
}
}
Assert.assertNotNull(matchingHistory);
String input1Id = null;
String input2Id = null;
for(final HistoryContents historyDataset :historyClient.showHistoryContents(matchingHistory.getId())) {
if(historyDataset.getName().equals("Input1")) {
input1Id = historyDataset.getId();
}
if(historyDataset.getName().equals("Input2")) {
input2Id = historyDataset.getId();
}
}
Workflow matchingWorkflow = null;
for(Workflow workflow : workflowsClient.getWorkflows()) {
if(workflow.getName().equals("TestWorkflow1")) {
matchingWorkflow = workflow;
}
}
final WorkflowDetails workflowDetails = workflowsClient.showWorkflow(matchingWorkflow.getId());
String workflowInput1Id = null;
String workflowInput2Id = null;
for(final Map.Entry<String, WorkflowInputDefinition> inputEntry : workflowDetails.getInputs().entrySet()) {
final String label = inputEntry.getValue().getLabel();
if(label.equals("WorkflowInput1")) {
workflowInput1Id = inputEntry.getKey();
}
if(label.equals("WorkflowInput2")) {
workflowInput2Id = inputEntry.getKey();
}
}
final WorkflowInputs inputs = new WorkflowInputs();
inputs.setDestination(new WorkflowInputs.ExistingHistory(matchingHistory.getId()));
inputs.setWorkflowId(matchingWorkflow.getId());
inputs.setInput(workflowInput1Id, new WorkflowInputs.WorkflowInput(input1Id, WorkflowInputs.InputSourceType.HDA));
inputs.setInput(workflowInput2Id, new WorkflowInputs.WorkflowInput(input2Id, WorkflowInputs.InputSourceType.HDA));
final WorkflowOutputs output = workflowsClient.runWorkflow(inputs);
System.out.println("Running workflow in history " + output.getHistoryId());
for(String outputId : output.getOutputIds()) {
System.out.println(" Workflow writing to output id " + outputId);
}The following code demonstrates creating a data library at four different levels of abstraction (these examples require an admin key):
// Most API methods have corresponding blend4j methods for dealing with
// both low-level request and parsed POJO responses. You can also use the method
// galaxyInstance.getWebResource() to access the low-level Jersey APIs directly.
final GalaxyInstance galaxyInstance = GalaxyInstanceFactory.get(url, apiKey);
final LibrariesClient librariesClient = galaxyInstance.getLibrariesClient();
// Highest level of abstraction, deal with POJO responses
final Library testLibrary1 = new Library("test1");
final Library persistedLibrary1 = librariesClient.createLibrary(testLibrary1);
// Deal with Jersey ClientResponse object in case want to check return status, etc...
final Library testLibrary2 = new Library("test2");
final ClientResponse response2 = librariesClient.createLibraryRequest(testLibrary2);
if(response2.getStatus() == 200) {
final Library persistedLibrary2 = response2.getEntity(Library.class);
// ...
}
// Use Jersey directly (with POJOs)
final WebResource webResource3 = galaxyInstance.getWebResource();
final Library testLibrary3 = new Library("test3");
final ClientResponse response3 = webResource3
.path("libraries")
.type(MediaType.APPLICATION_JSON)
.accept(MediaType.APPLICATION_JSON)
.post(ClientResponse.class, testLibrary3);
final Library persistedLibrary3 = response3.getEntity(Library.class);
// Use Jersey directly (no POJOs)
final WebResource webResource4 = galaxyInstance.getWebResource();
final ClientResponse response4 = webResource4
.path("libraries")
.type(MediaType.APPLICATION_JSON)
.accept(MediaType.APPLICATION_JSON)
.post(ClientResponse.class, "{\"name\": \"test4\"}");
final String jsonResponse4 = response4.getEntity(String.class);
System.out.println("JSON response is: " + jsonResponse4);This example demonstrates an admin process creating a private data library for a user and copying a directories contents to it. This example requires both an admin key and for use_remote_user to be enabled in the Galaxy instance's options file.
final GalaxyInstance galaxyInstance = GalaxyInstanceFactory.get(url, apiKey);
final String email = "[email protected]";
// Create data library
final Library library = new Library("Example library for " + email);
final LibrariesClient librariesClient = galaxyInstance.getLibrariesClient();
final Library persistedLibrary = librariesClient.createLibrary(library);
// Copy example directory into library
final FilesystemPathsLibraryUpload upload = new FilesystemPathsLibraryUpload();
final LibraryContent rootFolder = librariesClient.getRootFolder(persistedLibrary.getId());
upload.setFolderId(rootFolder.getId());
upload.setContent("test-data/variant_detection");
librariesClient.uploadFilesystemPathsRequest(persistedLibrary.getId(), upload);
// Obtain user object
User owner = null;
final UsersClient usersClient = galaxyInstance.getUsersClient();
for(final User user : usersClient.getUsers()) {
if(user.getEmail().equals(email)) {
owner = user;
break;
}
}
if(owner == null) {
// In order to create users like this - use_remote_user must be enabled
// in the Galaxy instance's universe_wsgi.ini options.
owner = usersClient.createUser(email);
}
// Obtain user role
Role ownersPrivateRole = null;
final RolesClient rolesClient = galaxyInstance.getRolesClient();
for(final Role role : rolesClient.getRoles()) {
if(role.getName().equals(email)) {
ownersPrivateRole = role;
break;
}
}
final String ownersPrivateRoleId = ownersPrivateRole.getId();
// Set data library permissions
final LibraryPermissions permissions = new LibraryPermissions();
permissions.getAccessInRoles().add(ownersPrivateRoleId);
permissions.getAddInRoles().add(ownersPrivateRoleId);
permissions.getManageInRoles().add(ownersPrivateRoleId);
permissions.getModifyInRoles().add(ownersPrivateRoleId);
librariesClient.setLibraryPermissions(persistedLibrary.getId(), permissions);These examples are taken from Examples.java. For more examples see these integration test cases.
Brad Chapman has put together a higher-level Clojure library - clj-blend, the source code of which can serve as additional examples of blend4j usage.
blend4j can be built with Apache Maven.
% git clone git://github.com/jmchilton/blend4j.git
% cd blend4j
% mvn compile
blend4j can be tested with Apache Maven.
% mvn test
In this default mode, blend4j will use the [galaxy-bootstrap][t1] library to download, configure, and run an instance of Galaxy for testing. blend4j can be tested against an existing instance using the following Java system properties.
% mvn -Dtest.api.key=<key> -Dtest.galaxy.instance=<url> test
The easiest way to obtain blend4j is using Maven Central.
For instance, if a project is using Maven, a blend4j dependency can be added by adding the following to the dependencies section of the project's pom.xml file.
<dependency>
<groupId>com.github.jmchilton.blend4j</groupId>
<artifactId>blend4j</artifactId>
<version>0.1-alpha-1</version>
</dependency>
Similar dependencies can be specified if using Ivy, Grape, Gradle, Buildr, or SBT. See mvnrepository for more details.
The latest blend4j jars can be downloaded from the Maven Central Repository.
The code is freely available under the Apache License Version 2.0.