Child pages
  • 2016-03-01 Meeting notes
Skip to end of metadata
Go to start of metadata

Date

Attendees

Updates

Tony

  • Continuing on OpenBEL API 0.6.0 release. The hold up is adding support for retrieving serialized BEL nanopubs (e.g. BEL Script, XBEL) from a translator using content negotiation.

Nick

  • Logging notification within BEL Manager.
  • Standing up an authenticated OpenBEL API.
  • RDF Transformations meeting.
  • Review PR #110.

Dexter

  • Prompted discussion on network representation that appeals to biologists and how OpenBEL can fill their needs. See Discussion below.

Discussion

The question was raised about how BKNs (BEL Knowledge Networks) were presented in OpenBEL Cytoscape tools. The bel-nav Cytoscape tool allows a user to incrementally load their BKN using the OpenBEL Web API. We represent a BKN in JSON Graph Format as well which allows for evidence as edge metadata. JSON Graph Format, with support for evidence metadata, can be imported into Cytoscape using JGFApp.

Dexter raised that NDEx will be soon be deployed on the desktops of many biologists along with other Cytoscape tools. Typically these biologists work with SIF (Simple interaction file) format for its simplicity. OpenBEL enriches SIF with functional terms that allows more specificity when needed. Additionally OpenBEL provides a supplementary model for biological assertions that may support a SIF interaction.

Here is example showing refinement of a basic interaction between two proteins. The intention is to convey that OpenBEL can present as SIF, but offer increasing specificity when traversing deeper into the BEL representation.

SIF basic interaction

AKT1 increases FOXO1

BEL formalism (refinement 1)

p(HGNC:AKT1) directlyIncreases p(HGNC:FOXO1)

  • subject: AKT1 protein
  • relationship: direct physical interaction relationship
  • object: FOXO1 protein
  • Valid SIF interaction

BEL formalism (refinement 2)

kin(p(HGNC:AKT1)) directlyIncreases p(HGNC:FOXO1,pmod(P,S,319))

    • subject: AKT1 acting as a kinase
    • relationship: direct physical interaction relationship
    • object: phosphoylated protein modification at serine, position 319
    • Valid SIF interaction

Literature support

    • First support

      • Citation: type: PubMed, name: Biochem J 2001 Mar 15 354(Pt 3) 605-12, id: 11237865

      • Support: FKHR, FKHRL1 and AFX are each phosphorylated by PKB at three residues in vitro, and, in co-transfection experiments with FKHR, these were shown to be Thr-24, Ser-256 and Ser-319 [9-14]. The half time for phosphorylation of each site was 2-5 min. In contrast, the mutation S256A completely prevented phosphorylation at Thr-24 and Ser-319 ... but the mutation T24A did not prevent phosphorylation at Ser-256 or Ser-319 ..., and the mutation S319A did not prevent the phosphorylation of Thr-24 and Ser-256...

      • Context: Species: 9606 CellStructure: Cell Nucleus

    • Second support
      • Citation: type: PubMed, name: EMBO J 2002 May 1 21(9) 2263-71, id: 11980723
      • Support: FKHR is phosphorylated by protein kinase B (PKB) at Thr24, Ser256 and Ser319 in response to growth factors, stimulating the nuclear exit and inactivation of this transcription factor.
      • Context: Species: 9606

 

In this example we initially present a low-fidelity SIF interaction between AKT1 and FOXO1. This raw form is easy for biologists to use but it is a generalization. We then refine our SIF interaction using BEL formalisms to increase the fidelity.

The discussion concluded with how to present BKNs to biologists in forms that suite their needs. Biologists working with models may want to see functional terms while a bioinformatician may want only SIF with gene symbol/id.

Some thoughts were elllucidated:

  • BKNs could allow for low- and high-fidelity transformations.
  • Transformations are the smallest functional units that can be composed together and/or sequentially applied.
  • Biologists should drive which transformations they need for their BKNs. Transformations are opt-in.
  • Content negotiation using HTTP could be a mechanism for enabling the client (e.g. biologists) to ask for what they would like their BKN to look like.

The discussion ended with an agreement to have some further discussion and put up a Google Doc to continue improving the idea.

Action

Tony

  • Finishing up BEL Nanopub serialization work within bel.rb/openbel-api.

Nick

  • RDF Transformations meeting.
  • Review PR #110.