Questions on BCO usage

[nleguillarme]

Hi, I am currently working on a schema for a knowledge base of traits and interactions.

I am currently trying to determine if BCO could be used as a basis for my ontology, and I have a few question concerning BCO usage :

• How can I link a 'taxonomic identification process' to a 'taxonomic assertion' ?
• How can I link an 'organismal entity' to an 'occurrence' ?
• Is there a way to express an assertion of the form ' identified as using BLAST on sequences from environmental DNA sample collected at at ' ?

My plan would be to extend BCO with new classes and properties for representing traits, interactions, measurement method, observation/occurrence context, etc.

Thank you.

[ramonawalls]

@nleguillarme Thanks for your inquiry. I hope that this paper: https://environmentalmicrobiome.biomedcentral.com/track/pdf/10.1186/s40793-015-0014-0 will answer some of your questions.

How can I link a 'taxonomic identification process' to a 'taxonomic assertion' ?

Figure 2 in the paper above directly addresses that issue, except we do not have a class for taxonomic assertion. Could you clarify what you mean by taxonomic assertion? We do have a class for identification assertion, and perhaps you could use this. In that case, the identification assertion (which is a data item) would be an output of a taxonomic identification process. If you need something different, we could look at a new class.

How can I link an 'organismal entity' to an 'occurrence' ?

We do not directly link dwc:Occurrence to any BCO classes. This is because Occurrence is intentionally semantically vague, making it very difficult to map to an ontology. In fact, some of the BCO developers, including me, would love to just get rid of Occurence in DwC, but we can't for legacy reasons. What I recommend instead is that for each instance of Occurence that you want to describe, you decide whether the evidence is an observation or a material sample or a published data source, then you make an instance of one of those classes to link to the organismal entity via a specimen collection process, an observing process, or some other planned process.

Is there a way to express an assertion of the form ' identified as using BLAST on sequences from environmental DNA sample collected at at ' ?

Question 3: This could be modeled in a couple of ways. One way would be to create an instance of DNA sequencing assay (http://purl.obolibrary.org/obo/OBI_0000626) (or RNA sequencing assay) which would have the environmental sample as input and sequence data as output, then the sequence data would be input to a BLAST process (which does not exist yet, but we could add to BCO or request it in EDAM). The BLAST process would have as with an instance of data item of the format BLAST (using http://edamontology.org/format_3331) would be input to an instance of taxonomic identification process and have as input your identification assertion.

My plan would be to extend BCO with new classes and properties for representing traits, interactions, measurement method, observation/occurrence context, etc.

I am happy to support requests for new terms, which you could either submit via the issue tracker or using a pull request. Please let me know if you think it would help to have a call to discuss your workflow.

[nleguillarme]

Thank you @ramonawalls for these clarifications.

1. I wrote 'taxonomic assertion' but I meant 'identification assertion', sorry for that. Figure 2 of the enclosed paper is what I was looking for, but with two issues. First, it seems that the only object property available to express the 'has output' relationship is 'has specified output' with domain 'planned process'. However, 'taxonomic identification process' is not a 'planned process', nor even a 'process'. Second, there is no class 'bco:taxon concept' in the ontology. I assume it now uses Taxon from DwC ?

2. I wanted to use Occurrence to provide some context to the organism observation, including location and date of the observation of course, but also sex and life stage of the organism, as suggested in the ETS standard vocabulary . It is not clear to me how I can do this with available classes and properties. In Figure 1 in the paper, a process is linked to a Location using the 'occurs in' property. However, 'process' is not defined as an 'occurent', nor is 'Location" defined as an 'independant continuant'. Though 'spatial region' in BFO is a subclass of 'independant continuant', it does not appear in bco.owl. For the second point (sex and life stage), I guess I can directly link the information to the 'organism'.

3. Though this is not priority, I am glad to see that this would be possible !

I am currently working on a figure in which I try to put things together and connect different ontologies relevant to my use case. I would be glad to share it and discuss it with you once I am done.

Thank you again for your support.

[ramonawalls]

1. I wrote 'taxonomic assertion' but I meant 'identification assertion', sorry for that. Figure 2 of the enclosed paper is what I was looking for, but with two issues. First, it seems that the only object property available to express the 'has output' relationship is 'has specified output' with domain 'planned process'. However, 'taxonomic identification process' is not a 'planned process', nor even a 'process'.

Hmm. Not being a planned process is a mistake. It looks like there is some problem with the OBI import that came up in the last release. I will fix that ASAP. Glad you noticed.

Second, there is no class 'bco:taxon concept' in the ontology. I assume it now uses Taxon from DwC ?

We are indeed currently using dwc:Taxon, rather than create our own class for taxonomic concept. I am a bit hesitant to get into the modeling of taxonomic concepts, and I was sort of hoping another ontology would cover it so we could import it. I'll have another look at what is out there. On the other hand, I am quite comfortable referring to a taxon as a collection of organisms, and I plan to make that term in the Population and Community Ontology (see #113)

2. I wanted to use Occurrence to provide some context to the organism observation, including location and date of the observation of course, but also sex and life stage of the organism, as suggested in the ETS standard vocabulary . It is not clear to me how I can do this with available classes and properties. In Figure 1 in the paper, a process is linked to a Location using the 'occurs in' property. However, 'process' is not defined as an 'occurent', nor is 'Location" defined as an 'independant continuant'. Though 'spatial region' in BFO is a subclass of 'independant continuant', it does not appear in bco.owl. For the second point (sex and life stage), I guess I can directly link the information to the 'organism'.

This may be related to the problem with taxonomic identification process. Process should indeed be an occurrent and Location should be an independent continuant. Another error for me to fix. A good lesson in how small changes to the import file can mess up the whole ontology!

3. Though this is not priority, I am glad to see that this would be possible !

As I mentioned, there are many ways it could be done. I am happy to discuss more.

I am currently working on a figure in which I try to put things together and connect different ontologies relevant to my use case. I would be glad to share it and discuss it with you once I am done.

I look forward to it.