Creating BioMoby WorkflowsIn Taverna

Mark WilkinsonEdward (Eddie) Kawas

iCAPTURE Centre, St. Paul’s HospitalVancouver, BC

Downloading Taverna

• Taverna can be obtained from:http://taverna.sourceforge.net

• Once at the site, click on Download

• Download the version appropriate for your operating system.

• Between releases, the Moby functionality may be updated and you can find instructions on how to acquire those updates from:http://biomoby.open-bio.org/index.php/moby-clients/taverna_plugin

Running

• For a more comprehensive guide on Running and using Taverna, please refer to http://taverna.sourceforge.net/usermanual/manual.html

• Assuming that you have downloaded and unzipped Taverna, you can run it by double-clicking on runme.bat (windows) or executing runme.sh (Unix/Linux/OS X)

• Taverna’s splash screen

• Once Taverna has loaded, you will see 3 windows:– Advanced Model Explorer

– Workflow Diagram– Available Services

• The Advanced model explorer is Taverna’s primary editor and allows you to load, save and edit any property of a workflow.

• Workflow diagram contains a read only graphical representation of your workflow.

• The Available services window lists all of the services available to a workflow designer.

• Under the node ‘Biomoby @ …’ Moby services and Moby data types are represented.

• The Object ontology is available as children of MOBY Objects node

• Services are sorted by service provider authority

• If you wish to use registries other than the default one, you can add a new Moby ‘Scavenger’ by choosing to ‘Add new Biomoby scavenger…’

• Enter the registry’s location and click okay.

Creating Workflows

• I have the workflow saved and would like to offer it for download.

• We will start by adding the Object ontology node Object to our workflow.

• The Advanced model explorer now shows that we have a processor called Object– Object has 3 input ports: id,

namespace and article name– Object has 1 output port:

mobyData• The Workflow diagram illustrates

our processor

• We can discover services that consume our data type, context click on ‘Object’ and choose ‘Moby Object Details’

• A window will pop up that tells you what services Object feeds into and is produced by

• Expanding the Feeds into node results in a list of service provider authorities• Expanding an authority, for example, bioinfo.icapture.ubc.ca, reveals a list of

services

• We will choose to add the service called ‘MOBYSHoundGetGenBankFasta’ to our workflow.

• A look at the state of our current workflow.

• And graphically.• The service consumes Object, with article name identifier,

and produces FASTA, with article name fasta.

• To discover more services, context click on the service that outputs the data type that you would like to discover consuming services for and choose Moby Service Details.

• The resultant window displays the services inputs and outputs.• There are also tool tips that show up when your mouse hovers over any

particular input or output that tells you what namespaces the data type is valid in

• Context clicking on an output reveals a menu with 3 options.

– A brief search for services that consume our datatype

– A semantic search for services that consume our datatype

– Adding a parser to the workflow that understands our datatype

• The result of choosing to add a parser for FASTA to our workflow.• The parser allows us to extract:

– The namespace and id from FASTA– The namespace and id from the child String– The textual content from the child String

• The result of choosing to conduct a brief search for services that consume FASTA

• We will add the service getDragonBlastText to our workflow by choosing ‘Add service -…’ from the context menu

• The current state of our workflow shown graphically.

• A more complex view of our workflow

• Finding services that consume NCBI_BLAST_Text starts by viewing the details of the service ‘getDragonBlastText’

• Conduct a brief search

• Add the service ‘parseBlastText’ to our workflow

• Our current workflow

• Workflow inputs are added by context clicking on Workflow inputs in the Advanced model explorer and choosing ‘Create New Input…’

• The result from adding 2 inputs:– Id– namespace

• The workflow input id will be connected to Object’s input port ‘id’

• Workflow after connecting the workflow input ‘id’

• The workflow input namespace will connect to Object’s input port ‘namespace’

• Workflow after connection the workflow inputs.

• Workflow outputs are added by context clicking on Workflow outputs in the Advanced model explorer and choosing ‘Create New Output…’

• The result from adding 2 workflow outputs:– moby_blast_ids– fasta_out

• The output moby_blast_ids will be connected to parseBlastText’s output port Object(Collection –’hit_ids’)

• The output fasta_out will be connected to Parse_Moby_Data_FASTA’s output port fasta_’content’

• To run the workflow, click on ‘Tools and Workflow Invocation’

• Choose ‘Run workflow’

• A prompt to add values to our 2 workflow inputs

• To add a value to the input ‘id’ click on id from the left pane and choose ‘New Input’

• Enter 656461 as the id

• Choose namespace from the left and click on ‘New Input’

• Enter NCBI_gi as the value for namespace• Once you are done, click on ‘Run Workflow’

• Our workflow in action

• Once the workflow is complete, we can examine the results of our workflow.

• A detailed report is available outlining what happened when and in what order.

• We can examine the intermediate inputs and output, as well as visualize our workflow.

• If we choose the Graph tab, our workflow is illustrated.

• Intermediate inputs allow us to examine what a service has accepted as input

• Similarly, Intermediate outputs allows us to examine the output from any particular service.

• Without the parser, FASTA is represented as a Moby message, fully enclosed in its wrapper.

• Non-moby services do not expect this kind of message

• Non-moby services expect the just the sequence and using the Parse_Moby_Data processor, we can extract just that

• Moby services can interact with the other services in Taverna.

• Let’s add a Soaplab service.

• We will choose a nucleic_restriction soaplab service called ‘restrict’

• Choose the restrict service and add it to the workflow.

• We will connect the output port fasta_’content’ from the service Parse_Moby_Data_FASTA to the input port ‘sequence_direct_data’ from the service restrict

• The result of our actions so far.• We will need to add another workflow

output to capture the output of restrict.

• Create an output called restrict_out

• Connect the output port ‘outfile’ from the service restrict to the workflow output restrict_out

• Once the connections have been made, run the workflow again using the same inputs.

• The workflow on the left has some extra services added to it.

– FASTA2HighestGenericSequenceObject from the authority bioinfo.icapture.ubc.ca

– runRepeatMasker from the authority genome.imim.es

– A Moby parser for the output DNASequence from runRepeatMasker.

– A workflow output Masked_Sequence

• Add them to your workflow

• The service runRepeatMasker is configurable, i.e. it consumes Secondary parameters.

• To edit these parameters, context click on the service and choose ‘Configure Moby Service’

• The name of the parameter is on the left and the value is on the right.

• Clicking on the Value will bring up a drop down menu, an input text field, or any other appropriate field depending on the parameter.

• The parameter species contains an enumerated list of possibilities.

• Select human.

• When you have made your selection, you may close the window.

• Let’s run the workflow

• We will run our workflow with a list– Click on id in the left pane and then click on New

Input twice

• Enter 656461 and 654321 as the ids• Enter NCBI_gi as the value for namespace• Our workflow will now run using each id with the single namespace

• Notice how the workflow is running with iterations. This is happening because the Enacter is performing a cross-product on the input

• You can still view intermediate inputs and outputs.• Using the queryIDs, you can track each invocation of a

moby service through the whole workflow

• Imagine now that you want to run the workflow using a FASTA sequence that you input yourself (without the gi identifier)

• To do this, context click on getDragonBlastText and choose Moby Service Details– Expand the Inputs node and context click on FASTA(‘sequence’)– Choose Add Datatype – FASTA(‘sequence’) to the workflow

• A FASTA datatype will be added to the workflow and the appropriate links created

• Notice the datatype FASTA on the left of the workflow– Since the datatype FASTA

hasa String, a String was also added to our workflow and the appropriate connection was made

• We will now have to add another workflow input and connect it to the String component of FASTA.

• A workflow input ‘sequence’ was added to the workflow and a connection was made from the workflow input to the input port ‘value’ of String.

• We also removed the link between MOBYSHoundGetGenBankFasta and getDragonBlastText by context clicking on the link in the Advanced model explorers’ Data links and choosing to remove the link

• Now when we choose to run our workflow, we will also have the chance to enter a FASTA sequence

• Go ahead an enter any FASTA sequence as the input to the workflow input ‘sequence’

• Run the workflow

• Any results can be saved by simply choosing to Save to disk– You will be prompted to enter a directory to save the results.– Each workflow output will be saved in a folder with the same name as a workflow

output and the contents of the folder will be the results• You can also choose Excel, which produces an Excel worksheet with

columns representing the workflow outputs and with rows that represent the actual data.