Dubious experimental annotations to behavior in human

[cmungall]

Background: given that BP is now defined as being about genetically programmed behaviors, this calls into question behavior annotations in humans, since in general most people would agree human behaviors are not genetically programmed (though "programming" may underpin many). See also obo-behavior/behavior-ontology#101

adult behavior

NLGN4X is annotated to 'adult behavior' (def: Behavior in a fully developed and mature organism)

http://amigo.geneontology.org/amigo/reference/PMID:12669065

I am not sure how an annotation to 'adult behavior' is useful, and how this is even justified by the paper which is about autism phenotypes. IMO This term should be a do_not_annotate

Other genes associated are NRXN2, NLGN3, SHANK1, SHANK3. I haven't
looked at these

Note that there are 160 direct and indirect annotations to 'adult
behavior' in human http://amigo.geneontology.org/amigo/term/GO:0030534

(48)	adult locomotory behavior
(31)	adult behavior
(27)	adult walking behavior
(20)	behavioral response to cocaine
(16)	behavioral response to ethanol
(11)	adult feeding behavior
(7)	behavioral response to nicotine

Some are IEA from mouse, some are IMP. all of these are suspect

feeding behavior

NPW is annotated to feeding behavior:

http://amigo.geneontology.org/amigo/reference/PMID:12959997

• I think 'regulation of feeding behavior' is more appropriate...
• But the paper is about rat not human!!! how did we get a human annotation?

Insulin is annotated to negative regulation of feeding behavior:

http://amigo.geneontology.org/amigo/reference/PMID:17957153

I'll concede this is justified but doesn't sit well. The same information can be communicated in a phenotype annotation. I think GO should be about mechanism, and should tell a more interesting story, e.g. about insulin's role in dopaminergic reward system.

SHANK1 Deletions in Males with Autism Spectrum Disorder

http://amigo.geneontology.org/amigo/reference/PMID:22503632

These seem like over-annotation based on phenotype. Certainly GO annotations don't communicate anything a phenotype annotation wouldn't.

There are some rat ISOs to the genes in this paper, including linking Shank1 to 'determination of affect'

id: GO:0050894
name: determination of affect
namespace: biological_process
def: "Any process in which an emotional response is associated with a particular sensory stimulation." [GOC:ai, GOC:dph, ISBN:0721662544]

Really? We can infer that a rat gene is involved rat feelings based
on the fact the human ortholog is implicated in autism? This seems suspect.

adult locomotory behavior

PAFAH1B1 involved in adult locomotory behavior

http://amigo.geneontology.org/amigo/reference/PMID:9063735

First, this is just a bizarre term to use for humans (def: Locomotory behavior in a fully developed and mature organism) -- how is locomotory behavior meaningfully different between a developed immature and mature human?

The paper links the PAFAH1B1 (LIS1) gene to lissencephaly, but doesn't actually have anything that could justify an IMP in it directly. But even if there were phenotypic observations, this should be a phenotype annotation.

motor behavior

id: GO:0061744
name: motor behavior
namespace: biological_process
def: "The specific neuromuscular movement of a single organism in response to external or internal stimuli."

DCTN1 is involved in motor behavior; IMP from:

http://amigo.geneontology.org/amigo/reference/PMID:18305234

This paper describes a mouse model! The methods and results are
entirely about mouse mutant phenotypes. How do we go from a mouse
model to a human IMP?

The paper does talk about human DCTN1 and motor neuron disease in the
intro/discussion. I think this should be reflected as TAS not IMP.

But really this information is best captured as a phenotype
annotation.

[RLovering]

Hi Chris

I am sure that the majority of these annotations are created by UCL, as we had a student curating genes associated with autism. I think at that time we were concerned about these annotations, but there were no other existing ontologies with active curation efforts that would enable the capture of this information. If HPO or DO is willing to use these papers to support the equivalent annotations that would be great. The identification of papers with appropriate evidence to support an annotation is time-consuming and it would be good if this time could be invested in a different resource if necessary rather than deleted immediately. I believe that this was agree in a GOC meeting many many years ago when behaviour annotations were first discussed.

Please note that the MOD annotations are not always based on the human phenotype associated with autism, and in fact the MOD data, which looks at the impact of a genetic mutation on behavioir, has often been used to support the suggestion that the ortholog in human that is suspected to be associated with autism is a good candidate.

Also note that to state that 'adult behaviour is not 'genetically programmed'' is under appreciating the data that exists. It implies that you consider humans as completely separate from our evolutionary past and suggests that the differences in behaviour of children and then adults is completely down to their upbringing. I don't think that all behavious are genetically programmed, with so many different metabolic processes influencing the biochemical homeostatis etc, however an example of genetic programming I can quickly think of is that mice are programmed to take risks otherwise mice would die of starvation, and I assume you would agree that testosterone does have an impact on behaviour in men, unless the man does not have the receptors required to respond to this hormone.

is this a use for our philosophy degrees???

Ruth

[RLovering]

Also FYI wrt your comment:

motor behavior

id: GO:0061744
name: motor behavior
namespace: biological_process
def: "The specific neuromuscular movement of a single organism in response to external or internal stimuli."

DCTN1 is involved in motor behavior; IMP from:

http://amigo.geneontology.org/amigo/reference/PMID:18305234

This paper describes a mouse model! The methods and results are
entirely about mouse mutant phenotypes. How do we go from a mouse
model to a human IMP?

The paper does talk about human DCTN1 and motor neuron disease in the
intro/discussion. I think this should be reflected as TAS not IMP.

The abstract states: mice expressing either wild-type or mutant human dynactin p150(Glued). It is common practice to curate the human proteins expressed in MOD systems, rather than the equivalent MOD ortholog. If you want to change this system of annotation then there is going to be a lot of annotations to revise.

As above I would be happy to discuss removing behaviour from GO but there needs to be an effective system available to replace it.

Ruth

[addiehl]

+1 to Ruth, "Also note that to state that 'adult behaviour is not 'genetically programmed'' is under appreciating the data that exists. It implies that you consider humans as completely separate from our evolutionary past and suggests that the differences in behaviour of children and then adults is completely down to their upbringing."

+1 to Ruth, "As above I would be happy to discuss removing behaviour from GO but there needs to be an effective system available to replace it."

There are clearly genetically based dispositions to a wide variety of behavior, many of which have been explored experimentally for the past century in both model organisms and humans. I think GO should continue to provide appropriate subclasses, particularly in the absence of a well-maintained alternative linked to an annotation system. When valid biological processes are omitted or removed from the the GO, the effective result many times is that gene product-linked data for the relevant domain will never annotated in a way that is widely available to the community, because of the reach of GO and the well developed annotation system around it. Some phenotype annotations cover some gaps, but not all.

[cmungall]

My intention was not to get philosophical. Of course I agree that behavior has some kind of genetic underpinning. But currently following the true path rule all behaviors in GO are entirely genetically programmed, based on the current definition of BP.

I should have stated from the outset that I'm not advocating immediate some kind of reactive immediate removal of all of these annotations. I would like to come up with a long term plan that allows us to better utilize dedicated phenotype databases and deliver higher value GO annotations that dissect the mechanisms of these phenotypes.

at that time we were concerned about these annotations, but there were no other existing ontologies with active curation efforts that would enable the capture of this information. If HPO or DO is willing to use these papers to support the equivalent annotations that would be great

Yes, HPO is set up to do this, we can discuss further.

Also note that to state that 'adult behaviour is not 'genetically programmed'' is under appreciating the data that exists. It implies that you consider humans as completely separate from our evolutionary past and suggests that the differences in behaviour of children and then adults is completely down to their upbringing

We are of course a product of evolution but still direct annotations to 'adult behavior' are odd, and I am not sure about direct annotations to 'adult locomotory behavior' as opposed to simply 'locomotory behavior'

[krchristie]

@cmungall - I'm not following your argument for why annotations to adult behavior are odd. I have made direct annotations to adult locomotory behavior for mice. I haven't looked back at of those papers, but there are definitely different behaviors in adult versus juvenile animals.

It might also be worth noting that there is also a term for larval locomotory behavior, suggesting that the distinction between adult and larval behaviors might be quite important in insects.

[ValWood]

@pgaudet
Maybe here an action could be to review the direct annotations to
(31) adult behavior (make this not for direct annotation)?
and weed out the ones which come from autism phenotypes.

The more specific descendent annotations are possibly OK (but maybe should not be propagated)?

[pgaudet]

@vanaukenk You said WB recently looked at behavior, is there any guidance your group can provide to help define these terms ?