chemical target phenotypes

[CuzickA]

These have been added to the 'single organism namespace' and are generic for both pathogen and hosts.

Do they fall under the 'cell level' or 'cell population level' phenotype or both?

For now, I have added under cell level

I have had a go at a logical definition please could I have some feedback @mah11

[mah11]

We have put analogous terms under 'cell population phenotype' in FYPO because the assays used to detect chemical sensitivity and resistance are generally done with cultures in liquid media or on plates; both methods evaluate the growth of a population of cells.

Depending on what is assayed in plant host-pathogen experiments, you may want to do the same or you may need to put some or all of the sensitivity/resistance phenotypes at the level of an organism (this would be the case if it's a host plant that is the chemical target). You can use 'cell level' if you do have any experiments that assess chemical sensitivity of a single cell, or if you're willing to overlook the nuances and infer sensitivity/resistance of a cell from what happens in an observed population.

[ValWood]

I would put them under the population branch, because we know this generally makes sense, and see if it needs to be tweaked.

[CuzickA]

Thanks for your input. I have now moved these terms under 'single organism cell population phenotype'.

Does this logical definition example seem reasonable?

I notice in FYPO that the 'increased resistance to chemical' terms are under 'cell population growth phenotype'. Should I consider moving these PHIPO chemistry terms to 'growth phenotypes'?

[CuzickA]

We will shortly have about 200 new Chemicals with CHEBI id to add to PHIPO. Currently each will need to be entered 4 times to be children of 'increased resistance to', 'increased sensitivity to', 'normal resistance to' and 'normal sensitivity to'. Is there a way that this can be automated as will be a lot of work to enter manually into Protege?

These 200 chemicals with also need to be entered into our PHI-eco conditions ontology.

[ValWood]

@mah11 do you know an easy way to 'automate this?'

[ValWood]

Does this logical definition example seem reasonable? also @mah does the logical def look OK? I'm still a bit confused about how they are constructed and need you to explain it to me again...

[CuzickA]

Talking with @martin2urban we wondered whether the AE penetrance would be available for these chemistry phenotypes? Can we record the concentration of chemical used in the assay?

[mah11]

Does this logical definition example seem reasonable?

I guess so. I'm not 100% sure about saying resistance or sensitivity inheres in an organism, as opposed to in a process, but it may be fine.

I notice in FYPO that the 'increased resistance to chemical' terms are under 'cell population growth phenotype'. Should I consider moving these PHIPO chemistry terms to 'growth phenotypes'?

It depends on what is affected. FYPO chemical sensitivity and resistance phenotypes are logically defined using GO:0072690 - which is the nearest thing we've found for the type of growth, although it's in the 'phase' branch of GO - and classified as cell population growth phenotypes because that's what people assay for (as noted above).

example - FYPO:0002578 resistance to hydroxyurea

'decreased sensitivity of a process'
and ('inheres in' some 'single-celled organism vegetative growth phase')
and (towards some hydroxyurea)

This exact pattern wouldn't work for any PHIPO terms that would apply to multicellular organisms, and even for us, it feels a bit hacky. This is on area (of many!) that I hope to review and improve when we go over FYPO with DavidOS and Nico.

Is there a way that this [adding lots of chemical-related terms] can be automated?

Sorry, I don't know of any way to do automated or bulk additions in Protege. Maybe it's worth asking how far off the successor to TermGenie is

[mah11]

we wondered whether the AE penetrance would be available for these chemistry phenotypes?

It would make no sense to put a penetrance extension on a fission yeast annotation using the FYPO equivalent of any of these terms, because we're already modeling them as phenotypes relevant at the level of a whole population of cells.

For PHIPO and its annotations, the same would apply if you also classify sensitivity/resistance as population phenotypes. If you don't, then you have to think about whether you'll actually have data that show what fraction of organisms in a population show the sensitivity or resistance; penetrance extensions are only meaningful when you have that kind of information.

Can we record the concentration of chemical used in the assay?

We haven't tried to capture this much detail in annotations using FYPO. That doesn't mean you can't or shouldn't, but it does mean we're not well equipped to teach you how; we don't have an already-working way to do it. As a stopgap you could put it in the comment field.

[ValWood]

Agreed, so in summary, penetrance only applies to ~cell level~ single organism phenotypes , if you annotate a specific observation (like misformed spindle) you can describe the cell level phenotype and say the % of cells this is observed in. Clearly this isn't possible at the population level. We have expressivity low, medium and high depending on the severity of the growth reduction.

We haven't recorded concentrations, it would be an added level of complication. You would understand why we don't want to record concentrations if you saw the sheer number of spot assays at different concentrations we need to curate.....

Part of our rationale was that we only want to provide the user that a particular sensitivity is observed in a particular mutant, and if that sensitivity is strong relative to other mutants (the biologically important observation). Not providing the exact concentration hasn't been an issue for our users because one would assume you would usually see a strong sensitivity phenotype at a similar concentration under the same conditions for a specific chemical.

It is a bit suboptimal for aggregating data across different publications, but even if you record the concentration the expressivity is still qualitative, so its a bit arbitrary without going to the paper to check the data.

Clearly, the concentration is useful information and you would need to know more about the precise conditions if you wanted to repeat the experiment, but we don't want to discourage people from reading the paper for the method details).

I also think the comments field is a good stopgap, or if you really wanted, you could maybe create some concentration ranges (i.e very broad orders of magnitude) in your conditions. We have done this for temperature ranges.

[mah11]

penetrance only applies to cell level phenotypes

Not quite. Penetrance applies to organism-level phenotypes, and not to population-level phenotypes. For FYPO annotations we can apply penetrance to cell-level phenotypes because our organism is a single cell.

[CuzickA]

Following on from a discussion on chemical phenotypes with @ValWood today It was suggested to remove the normal/abnormal groupings and use 'resistance to' or 'sensitive to' to follow FYPO. The PHIPO slim terms have historically used different definitions for these terms I think @ValWood suggested that we could use the AE 'has expressivity' to differentiate between normal wildtype levels of resistance/sensitivity to a chemistry and increased resistance/sensitivity to a chemistry.

A normal phenotype to chemistry could be captured as 'normal growth' with condition +Benomyl

This is what phipo could look like if we decide to go ahead with this structure.

We would like to put some thought into getting this right before generating the new terms for our list of 200+ chemicals

Just to note that @martin2urban also supports the suggested changes above

[ValWood]

An additional advantage is that it is then easy to retreive all genes in a species that display sensitivity or resistance. This would not be easy if you arbitrarily defined normal based on a specific experiment.

[CuzickA]

How would we go about creating a logical definition for PHIPO 'resistance to benomyl' which would now be under the 'population phenotypes'?

In FYPO it looks like this

@mah has commented on this above in the ticket and suggests it may not be suitable for PHIPO. Any further suggestions on whether we could use this 'decreased sensitivity of a process' structure or any other ideas would be most welcome thanks.

[mah11]

It will probably be worth watching to see what pattern(s) eventually emerge from the uPheno work. In the meantime, I don't know what would be better from PATO than 'decreased (or increased) sensitivity of a process'. The part I think might not work for PHIPO is the process that we use for FYPO sensitivity/resistance - mostly 'single-celled organism vegetative growth phase' (GO:0072690). That wouldn't work for any multicellular organisms.

[ValWood]

@CuzickA what is measured when sensitivity is assayed?

This phenotype must exist in other ontologies for multicelluar species? Might be a good first question for the slack channel... Especially is you say how FYPO did it and why that isn't appropriate for PHIPO.

'single-celled organism vegetative growth phase' is a slightly odd term in GO and there is no multicellular equivalent.

[CuzickA]

For 'resistance to benomyl' would this logical definition work? @mah11

'decreased sensitivity of a process' and (inheres_in some growth) and (towards some benomyl)

Growth is

Using @ValWood nifty basket trick on quickGO here is where growth sits. I don't think we really want to be capturing single-celled and multicellular organism phenotypes separately do we?

[CuzickA]

I am still a bit unsure about the parents for the chemistry terms (single species namespace)- whether they are to be grouped under 'cell population phenotypes' or something more like 'organism growth phenotypes'? I currently have

I think I will need 'organism' to build the logical definitions but GO does not have 'organism'

Other ontologies including Monarch (MONDO) have 'organism'

Should we import 'organism' from one of these other ontologies?

[ValWood]

@mah ? It seems a bit overkill to use an ontology for a single term that should be defined somewhere generically.

Isn't there anything suitable in PATO could a sensitivity be described as an organismal quality?

Your suggestion 'decreased sensitivity of a process' and (inheres_in some growth) and (towards some benomyl)

might work. The term "growth" can apply to multicellular of single celled organimss. Here growth means an increase in size though. The term in the phase branch is referring to a "life cycle stage"

I would probably leave the logical def until the POTATO meeting, or ask on the OBo list...

[ValWood]

I can't find anything in upheno for chemical sensitivities or resistance. It would be useful if there was an easy way to search only ontologies which have logical defs. Anyone?

[mah11]

For 'resistance to benomyl' would this logical definition work?

'decreased sensitivity of a process'
and (inheres_in some growth)
and (towards some benomyl)

It really depends on whether 'growth' (i.e. all or part of the organism increasing mass) is the process that would be affected by benomyl if the resistance didn't happen. I don't know if that's always the case (or indeed if it's ever the case, but it would have to be 'always' to build that into the ontology).

I am still a bit unsure about the parents for the chemistry terms (single species namespace)- whether they are to be grouped under 'cell population phenotypes' or something more like 'organism growth phenotypes'?

I'm pretty sure you wouldn't use 'cell population' or 'organism population' as FYPO does - that works for FYPO for two reasons, neither of which apply to your pathogen-host interactions (as I understand them). One is that fission yeast is a single-cell organism, so a cell population is an organism population; the other is that chemical sensitivity and resistance are assayed by evaluating growth of a cell/organism population. The chemical does or doesn't affect an individual cell's ability to grow and divide, but as a rule assays aren't set up to try to detect that.

The FYPO logical def is a bit of a hack anyway, for reasons that would be off on yet another tangent ...

I think I will need 'organism' to build the logical definitions but GO does not have 'organism'

'Organism' is out of scope for GO.

Should we import 'organism' from one of these other ontologies?

I tend to agree with Val - it's probably overkill at least for the moment.

[CuzickA]

Just to note that our lists of precomposed anti-infective chemistry terms have now been added to the 'single species phenotype' branch of phipo with text definitions but NOT logical definitions. Chemicals have also been added to phi-eco.

Egs for the 'single species phenotype' branch of phipo

[CuzickA]

Closing ticket for now. Logical definitions should be captured with the upheno patterns in the future.