Locked History Actions

Diff for "PropheticGrangerUseCase"

Differences between revisions 21 and 22
Revision 21 as of 2015-02-19 16:14:07
Size: 3453
Editor: OriolGuitart
Comment:
Revision 22 as of 2015-02-19 16:15:13
Size: 3473
Editor: OriolGuitart
Comment:
Deletions are marked like this. Additions are marked like this.
Line 11: Line 11:
 1. Translate provided HUGO column to real HUGO IDs. Since there might be several HUGO IDs for a protein, the HUGO IDs need to be inserted in the same cell separated  by a comma.  1. Translate provided HUGO column to real HUGO IDs. Since there might be several HUGO IDs for a protein, the HUGO IDs for the same protein need to be inserted in the same cell separated by a comma.

Step-By-Step Use Case

Here, we present a sample use case to show all possibilities that new PropheticGranger app provides. We provide real breast cancer data from DREAM 2008 challenge to illustrate all required steps to use PropheticGranger app to infer a network from this data.

This use case dataset was generated using Reverse Phase Protein Array (RPPA) quantitative proteomics technology and it focus on ~45 phosphoproteins (proteins phosphorylated at specific sites). Time-course data were acquired under 8 ligand stimuli and inhibition of network nodes by one of 3 inhibitors or DMSO vehicle control (cells were serum-starved and pre-treated with inhibitor prior to ligand stimulation). The experiment was carried out on 4 breast cancer cell lines, with abundance of ~45 phosphoproteins measured at 7 time points post-stimulus. Data provided are normalized protein abundance measurements on a linear scale. Figure below shows a representaion of the used data.

Dream Data

The original data can be downloaded here. However, this data needs to be modified to be used correctly by the PropheticGranger app. These are the steps needed to get a dataset ready to be used.

  1. Transpose original data so that rows correpond to proteins and columns to observations.
  2. Translate provided HUGO column to real HUGO IDs. Since there might be several HUGO IDs for a protein, the HUGO IDs for the same protein need to be inserted in the same cell separated by a comma.
  3. Remove Cell Line column (row in the transposed version).
  4. Merge inhibitor, stimulus and time point to form the new column names. The use of multidimensional data in PropheticGranger app is expained in detail here.

The new processed data is available here and it is now ready to be loaded to Cytoscape. Therefore, we are ready to start using PropheticGranger app to infer a network from the breast cancer cell data. Next lines explain step by step how to infer a network from this data.

  1. Install and start Cytoscape 3.2.0 or above
  2. Install PropheticGranger from Apps > App Manager > Install from File.

  3. Select PropheticGranger jar file

  4. Select File > Import > Table > File...

  5. Follow this loading instructions

  6. Click on Cyni Toolbox in Cytoscape Control Panel

  7. Select PropheticGranger Algorithm as the Inference Algorithm

  8. Select previous loaded table as the Table Data

  9. Select Use Human Pathway Commons database prior data in the Prior Data Definition box.

  10. Select HUGO ID as the Select column containing HUGO Ids.

  11. If needed, select all elements in Data Attributes selection box.

  12. Click Apply to start inferring networks from data

The process of infering a network from the provided data takes a few minutes. When the process is finished, a new network is generated for each stimulus on the provided data. Below you can see how these inferred networks look. Edge table for each network contains the score that PropheticGranger method has assigned to each edge.

Inference results