Ontology meeting 2012-03-28

From GO Wiki
Jump to: navigation, search

1 hour call

DISCUSSION 1: UPDATE ON GO-CHEBI ALIGNMENT

What are our next steps now we've switched to svn? Where are we on the paper?

DISCUSSION 2: PEPTIDE v POLYPEPTIDE v PROTEIN

How can we better define peptide vs protein in GO? At the moment, we're not consistent. For most terms (e.g. peptide transport, protein transport) they're siblings. Does a protein have to have tertiary/quaternary structure?

CHEBI has:

peptide ; CHEBI:16670
Amide derived from two or more amino carboxylic acid molecules (the same or different) by formation of a covalent bond from  the carbonyl carbon of  one to the nitrogen atom of another with formal loss of water. The term is usually applied to structures formed from α-amino acids, but it includes those derived from any amino carboxylic acid.
polypeptide ; CHEBI:15841 
A peptide containing ten or more amino acid residues.
peptide ; CHEBI:16670
--[isa]polypeptide ; CHEBI:15841

PROTEIN ONTOLOGY has:

amino acid chain ; PR:000018263
--[isa]protein ; PR:000000001
amino acid chain ; PR:000018263
A polymer of amino acids linked by peptide bonds.
Narrow synonym: peptide
Narrow synonym: polypeptide
protein ; PR:000000001
An amino acid chain that is produced de novo by ribosome-mediated translation of a genetically-encoded mRNA.


Stems from a SF item from Val: http://sourceforge.net/tracker/?func=detail&aid=3511488&group_id=36855&atid=440764


DISCUSSION 3: Formally specifying regulation propagation rules

See email thread.

DISCUSSION 4: Generating GO slims for website

Are we going to use Oort to generate these? Progress?



MINUTES: BECKY

PRESENT:

  • Chris
  • David
  • Tanya
  • Jane
  • Rebecca
  • Paola
  • Karen (for peptide and ChEBI discussion)
  • Harold (for peptide and ChEBI discussion)


MINUTES: UPDATE ON GO-CHEBI ALIGNMENT

The situation has changed slightly from the UCL discussion because we're making a new version of ChEBI (uchebi, u=union), which contains novel uchebi IDs (union IDs for conjugate acids and bases). E.g.

[Term]
id: UCHEBI:30769
name: citric acid or conjugate
alt_id: CHEBI:16947
alt_id: CHEBI:30769
alt_id: CHEBI:35804
alt_id: CHEBI:35808
subset: uchebi
synonym: "citrate(1-)" NARROW []
synonym: "citrate(2-)" NARROW []
synonym: "citrate(3-)" NARROW []
synonym: "citric acid" NARROW []
is_a: CHEBI:27093 ! tricarboxylic acid
  • These union terms are automatically generated and formally defined (so we shouldn't need to manually edit this file)
  • Heiko's script uses uchebi to build an is_a hierarchy with all conjugate acids and bases merged. This should reflect the structure of GO.

  • AI1: David and Tanya to cast their eyes over uchebi.owl or uchebi.obo to check it makes sense.
  • AI2: Chris to perform an alignment of GO with ChEBI, and generate a diff of the changes to GO that this would cause. Tanya, David and Jane to divide up and look at the diff next week, and check it makes sense (similar to regulates sanity check).
  • AI3: Chris to tweak CHEBI paper to explain how we're handling these unions, then can get it sent off.


MINUTES: PROTEIN VS PEPTIDE

  • It's not just GO that's confused (not sure if that's a good or a bad thing, considering we'd like a reference point).
  • The PRO set up was designed to distinguish between peptides synthesised by the ribosome and those non-ribosomally encoded (e.g. glutathione). For GO however, it probably doesn't matter to us how it's derived because we're just interested in its structure.
  • Additional problem with relationships between peptides and proteins, is that many proteins are chopped up to form peptides.
  • Tertiary/quaternary structure can NOT be used to define a protein.
  • AI: Harold responsible for hassling ChEBI to get this issue resolved, and produce a consensus between PRO and CHEBI. Should mail the PRO and CHEBI mailing lists to keep everyone informed.


MINUTES: PROPAGATION OF REGULATION

  • SUMMARY OF VAL'S ISSUE: Should annotations be propagated along the regulates relationship?
  • To know when it's involved in a process directly, and when it regulates a process, you need to know the function of the GP. If the function is part of that process, then your GP is part of the process, not regulating it. For many GPs though, we don't know the MF; we just know the process that's affected (and we don't know whether it's part of the process or regulating the process). Formerly, if we didn't know whether the GP was part of the process or was regulating the process, we'd annotate to the process.
  • Chris suggested the possibility of being able to annotate to 'process OR regulates process.... we don't know which'.
  • David suggested annotating to a function, and putting 'part_of process x' in column 16. or 'regulation of process x' in column 16.
  • Tanya suggested analysing the existing functions for GPs, and using them to clean up the existing process vs regulates_a_process annotations.

AI: Need to present this task to the PIs to see if they think a clean-up is a priority. AI: Ran out of time, so need to come back to the specifics of this on the next call, with exactly what to present to PIs.