ChIP annotation? DNA sequence-specific chromatin binding

[gocentral]

The chromatin immunoprecipitation assay (ChIP) is typically used to determine what proteins “bind” to what sequence of DNA. The limitation of the assay really restricts the data to what proteins “associate” with a particular sequence of DNA. In the usual ChIP assay the proteins are crosslinked to proteins as well as some of the proteins being crosslinked to DNA. This leads to vagueness of conclusions, because after the immunoprecipitation step, the DNA is separated from the protein and it really can’t be determined which protein was directly or indirectly bound to DNA. Similarly, in native ChIP when no crosslinking is used, the end result is also vague on what protein was directly or indirectly bound to DNA. Therefore to deal with the ambiquity, I would like to see the child term “DNA sequence-specific chromatin binding” added as a child of “chromatin binding” (GO:0003682). The definition could be “Interacting selectively and non-covalently with DNA sequence-specific chromatin, the network of fibers of DNA, protein, and sometimes RNA, that make up the chromosomes of the eukaryotic nucleus during interphase.” A parallel term of “DNA sequence-specific nucleosome binding” could be added as a child of “nucleosome binding” (GO:0031491). PMID:20816704, PMID:10075891

Reported by: slaulederkind

Original Ticket: geneontology/annotation-issues/944

[gocentral]

I think we decided at some point that you could not use ChIP to annotate to DNA binding because the binding is often indirect. I was looking for this documentation recently on the GO site to point our new curator to it, but I couldn't find it anywhere....anyway I wanted to flag that if we do this, it would be a change of policy and would require wider broadcasting....and the old documentation if it still exists would need updating. Val

Original comment by: ValWood

[gocentral]

Hi Val, I don't think this would be a change in policy. On the contrary it gives an alternate way to annotate from ChIP data that avoids any claim of protein-DNA binding. Stan

Original comment by: slaulederkind

[gocentral]

I'm going to move this to the annotation tracker to get an answer to the question about ChIP annotation.

Emily/Rama - please move it back to ontology tracker if you think we should annotate this.

Original comment by: jl242

[gocentral]

• labels: 390929 -->
• summary: DNA sequence-specific chromatin binding --> ChIP annotation? DNA sequence-specific chromatin binding

Original comment by: jl242

[gocentral]

The current documentation on use of ChIP experiments is here; http://wiki.geneontology.org/index.php/Annotation\_consistency\_:\_ChIP\_experiments It's not really guidelines, more a record of communications with IntAct curators - but it could be made into guidelines.

I asked Sandra to comment on Stan's suggested term and she thinks it sounds reasonable.

She has also given me some more indication of when IntAct annotate using ChIP experiments, which I have added to the wiki page. It sounds like they are very cautious and at most can only claim an 'association' which may be direct or indirect.

Hope this helps. Rachael.

Original comment by: rachhuntley

[gocentral]

Stan, If it is not clear that the protein you are annotating is bound directly or indirectly to DNA, then annotating to a term like DNA sequence-specific chromatin binding would be misleading I feel. Unless I am misunderstanding your question. So I wouldn't add these terms.

Rama

Original comment by: rbalakri

[gocentral]

Hi Rama, The idea is to have something more specific than "chromatin binding" for annotating data from ChIP experiments. Maybe you could phrase it another way, but "DNA sequence-specific chromatin binding" to me means a protein is binding to a specific chunk of chromatin that is dependent on the DNA sequence in that chunk of chromatin. Stan

Original comment by: slaulederkind

[gocentral]

Hi Stan,

I think that the problem with trying to create a Function term in the binding tree to annotate ChIP experiments is that we have been saying for quite some time that you should only annotate to "binding" terms if you know the binding is direct. Thus, as others have pointed out previously, use of ChIP experiments is inherently problematic for use of any of the "binding" terms in the MF ontology.

If you annotate proteins from ChIP experiments to a term called "DNA sequence-specific chromatin binding" which would presumably have a relationship to the term "sequence specific DNA binding" then you might get non-DNA binding proteins in enrichment analyses for "sequence-specific DNA binding". It isn't any better if you follow up a putative relationship up the "chromatin binding" side since, again, you don't actually know that the protein is binding to chromatin either; it might be binding to something else that actually binds chromatin directly. Thus, we have been saying for some time that ChIP experiments really cannot be used as evidence for MF binding terms.

So, the recommendation we have been making if you wish to annotate these is that you can say that protein x "colocalizes with" the CC term "chromatin", or perhaps with one of its child terms. At one of the GOC meetings in Cambridge a few years ago, 2009 perhaps, there was a lot of discussion about this and some ideas about whether we could put terms in GO that represented regions like "promoter", "5' region of gene", "3' region of gene", so that you could annotate what is often the real goal of ChIP and indicate the type of region the protein is annotated to. However, in discussions with Karen Eilbeck (of the Sequence Ontology), the conclusion was that this is probably better done via some sort of column 16 info using a SO term.

-Karen

Original comment by: krchristie

[gocentral]

• status: open --> closed-works-for-me
• Group: --> UniProt

Original comment by: rachhuntley

[gocentral]

I have updated the GO wiki page http://wiki.geneontology.org/index.php/Annotation_consistency_:_ChIP_experiments with the following:

Update 5 Nov 2014 Discussed via [SourceForge] In summary: An acceptable annotation based on ChIP is protein x "chromatin" (or a child term) and then include Sequence Ontology (SO) term in the annotation extension field.

Note that the definition of chromatin is: The ordered and organized complex of DNA, protein, and sometimes RNA, that forms the chromosome therefore it is implied that any protein identified by ChIP is part of the chromatin rather than binding the chromatin.

Question: Based on discussions at the Barcelona GOC meeting and comments from IntAct above can we say that if the protein is known to be able to bind DNA (based on it's structure) and the authors are therefore doing a ChIP to confirm what promoters the DNA binding protein binds to that it is acceptable to annotate to GO:0044212 transcription regulatory region DNA binding (or child term) and then include the Gene ID for the gene associated with that promoter in the annotation extension field?

Ill also email all the people who discussed this above

Ruth

Original comment by: RLovering