myExperiment.org - Workflows - Protein_transmembrane_prediction (Hamish McWilliam) [Taverna 1 Workflow]

Workflow Entry: Protein_transmembrane_prediction

Created at: 01/06/08 @ 17:49:32 Last updated: 26/10/08 @ 21:16:10

Version 2 (latest) (of 2)

Version created on: 26/10/08 @ 21:16:10 by: Hamish McWilliam | Revision comments Expand

Title: Protein_transmembrane_prediction

Type: Taverna 1

Preview

(Click on the image to get the full size)

Download Scalable Diagram (SVG)

Description

Transmembrane and signal peptide prediction using three methods: 1. EMBOSS tmap with a single sequence. Uses Soaplab tmap. 2. Phobius. Uses EBI’s WSPhobius web service. 3. TMHMM and SignalP. Uses the TMHMM and SignalP methods of InterProScan via the EBI’s WSInterProScan service.

The results of the three methods are converted into GFF format and collated.

Download

Download Workflow File/Package (SCUFL)

Run

Run this Workflow in the Taverna Workbench...

Option 1:

Launch in Taverna

Note: you need to have both the WHIP Launcher and the Taverna myExperiment/WHIP plugin installed on your machine for this to work. See here for information.

Option 2:

Copy and paste this link into File > 'Open workflow location...'
http://www.myexperiment.org/workflows/216/download?version=2
[ More Info Expand ]

Workflow Components

Inputs (2)

Name	Description
Email	User e-mail address.
Sequence_or_ID	Sequence to analyse. Either the actual sequence (fasta format recommended) or an entry identifier in database:identifer format (e.g. uniprot:LPHN2_RAT).

Processors (4)

Name	Type	Description
EBI_Phobius	workflow	Predict transmembrane domains and signal peptide using Phobius.
EBI_InterProScan_tmhmm_signalp	workflow	Predict transmembrane regions using TMHMM and signal peptide using SignalP.
tmap_single_seq	workflow	Transmembrane region prediction using EMBOSS tmap.
Merge_GFF	beanshell	Collate GFF output from the prediction methods.

Outputs (4)

Name	Description
Phobius_output	Predictions from Phobius.
InterProScan_output	InterProScan methods TMHMM and SignalP predictions.
tmap_output	EMBOSS tmap prediction, from single sequence input.
GFF_output	Collated feature predictions in GFF format.

Links (12)

Sink	Source
EBI_InterProScan_tmhmm_signalp:Email_address	Email
EBI_Phobius:Email	Email
EBI_InterProScan_tmhmm_signalp:Sequence_or_ID	Sequence_or_ID
EBI_Phobius:Sequence_or_ID	Sequence_or_ID
tmap_single_seq:Sequence_or_ID	Sequence_or_ID
Merge_GFF:gff_input2	EBI_InterProScan_tmhmm_signalp:GFF_result
Merge_GFF:gff_input1	EBI_Phobius:Phobius_GFF
Merge_GFF:gff_input3	tmap_single_seq:tmap_GFF
InterProScan_output	EBI_InterProScan_tmhmm_signalp:InterProScan_text_result
Phobius_output	EBI_Phobius:Phobius_output
GFF_output	Merge_GFF:gff_output
tmap_output	tmap_single_seq:tmap_report

Coordinations (0)

Citations (0)

None

Version History

Earliest Version:
[1] - Protein_transmembrane_prediction

Created on: Sunday 01 June 2008 @ 17:49:30 (BST)

Created by: Hamish McWilliam

Revision comments:

None

Latest Version:
[2] - Protein_transmembrane_prediction

Created on: Sunday 26 October 2008 @ 21:16:10 (GMT)

Created by: Hamish McWilliam

Revision comments:

Update for changes in the WSInterProScan web service.

Comments (3)

View Timeline

Log in to make a comment

Niall Haslam

Any chance of getting the subworkflows for this as well. Couldn't find them online and I want the polling part. i.e. For loads of the EBI jobs, you have to write beanshells to check if the job is done. It would nice to see a write up on how to create workflows for this.

Thanks,

Niall

Hamish McWilliam

Hi Niall. The subworkflows are included. If you download this workflow, load it into Taverna and navigate to the nexted workflows you can save them indpendantly using the normal save functionality.

The general process required to poll for service results is explained on the EBI website (see http://www.ebi.ac.uk/Tools/webservices/tutorials/workflow/taverna). In brief it uses Taverna's retry mechanism and a nested workflow which fails if the job has not completed.

Giovanni Dall'Olio

Great!

With which proteins have you tried this workflow? Do you have an estimate for its precision/accuracy?

There are other protein structure preditors that merge results from various sources, and combine them, to obtain a better prediction.
Is this the same approach you are using in this protocol?
e.g. I know of this predictor (http://pongo.biocomp.unibo.it/), but it is only for membrane proteins structure. Is it something similar?

How are the gff results merged at the end of the workflow? What does Gff_output contains, exactly?