Menu
▾
▴
Re: [Obo-discuss] [Obo-taxonomy] [Phenoscape] Proposal for a rank ontology (vocabulary)
|
From: Pankaj J. <pj...@co...> - 2009-06-08 20:57:08
|
Hilmar, That's why in the first place I suggested of using the ranks as relationships, instead of taxon-X | has_rank | rank-Y (e.g. O. sativa | has_rank | species). With my proposal one can build the phylogenomic/a taxonomy tree in its simplest of the form. See more below. Alternatively if we go by current suggestion of taxon-X has_rank rank-Y strategy, one will end up having ranks as classes, but children of those classes will be nothing but a long flat list of terms, e.g. imagine more than million species terms as children of rank_species, which may not resolve any of the following questions I had. Regarding the multiple phylogenetic trees, there is a diverse array of trees and not many consensus ones. Most of the studies are done using certain molecular and morphological markers which may or may not represent the structure and organization of the whole genomes. If the whole genomes are sequenced** and aligned, we may see a different phylogeny compared to certain loci which may have their won localized phylogeny depending on how actively evolving is that region of the genome in a given species. So if we traverse the whole genome based phylogeny we may see it breaking it a certain points compared to the whole genome. I am sure as we sequence more genomes this will be the method of choice. Conventional taxonomy, on the other hand followed by users and text books use Linnaean taxonomy that is taught from K-12 and even in colleges and is widely used. The question comes to a point where we see diverse sources of taxonomy and many using their own version as we have seen people mentioning it in this thread Not to forget 'no_rank in NCBI's taxonomy version. Besides ontology use the consensus can be built among various versions, using the similar strategy described by Yap et al (Genetics, Vol. 165, 2235-2247, December 2003 http://www.genetics.org/cgi/content/abstract/165/4/2235) for genetic maps, where we see a common problem of diverse position/order (top to bottom) on a given genetic map. Analogous to taxonomy which is one single genetic map with taxons ordered from top to bottom. Let me know if it helps. Pankaj ** Whole genome sequencing is not necessarily a representation of the whole species, instead a version of its genome based on the strain that was sequenced. A strain may have a gene that is knocked out / truncated or even a null allele, it may still put the phylogenoy out of place, and will not be a true representation in the tree Hilmar Lapp wrote: > Pankaj - these are great examples. How do you expect a taxon rank > ontology to help with or be applied towards those unless you include > phylogenetic definitions (which I believe would be highly contentious > as these would necessarily be hypotheses, not what we know about > reality). > > -hilmar > > On Jun 5, 2009, at 1:25 PM, Pankaj Jaiswal (OSU) wrote: > > >> Here's the tough part with a couple of examples. >> >> -What are the morphological and developmental landmarks that e.g. >> various Oryza species ( sativa, galberrima, etc) and its siblings >> share >> and/or differ on. >> - I presume the shared characters can them be the landmarks for higher >> order genus or any parents in the tree. If we go by the current >> practices of annotations the parent terms will obviously acquire both >> the common and uncommon features. However, the uncommon features will >> not be applicable intransitively since not all the children will have >> all those feature. >> >> -If a (set of) gene(s) displays phenotype/function/expression for >> example in O. sativa species, what's the likelihood of the orthologous >> gene displaying the same phenotype/function/expression in sibling >> species or a related species in a phylogenetic tree but coming from a >> different genus of the common tribe/phyla. >> >> -How does these set of orthologous genes evolved in different species >> (or even accessions of the species) and contribute to genetic >> diversity, >> response to environment/envO (biotic and abiotic) and georeference >> site >> (GAZ), loss/gain of function and paralogous pairs that may also >> involve >> ploidy or genome duplication. >> >> - some may also want to study the order of genes in syntenic region >> between the two species >> >> - extend it to study of regulatory cis and trans elements of the >> orthologous genes. >> >> Pankaj >> >> >> Erick Antezana wrote: >> >>> Hi Pankaj, >>> >>> could you elaborate a bit more on the "reasoning" scenarios you >>> mentioned? (competency questions, use cases,...) >>> >>> thanks, >>> Erick >>> >>> 2009/6/1 Pankaj Jaiswal (OSU) <jai...@sc... >>> <mailto:jai...@sc...>> >>> >>> Peter, >>> >>> I agree with your thought on not to avoid the taxonomic >>> inheritance. I >>> am sure Todd Vision is following this as he is involved in some >>> of the >>> similar reasoning projects on the iPlant. There is more >>> involvement of >>> iPlant in other projects e.g on phenology, where we need these >>> reasonings to be built and I am participating in them. In the >>> meantime >>> as I have seen, often more granular terms in taxonomy edges >>> belonging to >>> rank species and below are most detailed and are somewhat fixed. >>> Whereas >>> their parentage nodes may change as we learn more about the >>> biological >>> systems and phylogenomics using molecular genetic and other >>> approaches. >>> Your transitive approach seems great as a wholesome approach, >>> however, >>> often we can say with confidence in an intransitive fashion what >>> do O. >>> sativa and its siblings contribute to Oryzeae to make it unique as >>> compared to what is unique about Oryzae that is always inherited >>> transitively in all the siblings for example Oryza, Zizania, and >>> Leersia. We know everything is not inherited transitively in >>> this case. >>> >>> Pankaj >>> >>> Peter Midford wrote: >>> >>>> Pankaj, >>>> I'm happy to see a Botanical perspective on this, >>>> >>> since it >>> >>>> would be best to come up with a solution that works for everyone. >>>> My >>>> two questions about treating ranks as relationship types are: i) >>>> >>> how to >>> >>>> define them to capture the intended meaning - making them simple >>>> subrelations of is_a, with no differentia won't go over with Barry I >>>> expect; ii) how to reason with them. I don't know if you do any >>>> reasoning using your taxonomy, but in Phenoscape we decided that >>>> >>> trying >>> >>>> to avoid using taxonomic inheritance is impractical. I doubt we >>>> >>> are the >>> >>>> only project that has been led back to taxonomic inheritance for >>>> practical reasons. So, if you want to inherit properties from >>>> >>> Oryzeae >>> >>>> down to Oryza sativa, than rather than saying that is_a is >>>> >>> transitive, >>> >>>> you would need to define inheritance for compositions (e.g., >>>> is_species_of compose is_genus_of) to infer that O. sativa >>>> >>> inherits from >>> >>>> Oryzeae. >>>> >>>> >>>> I'd be happy to see solutions to these, as it sounds like an >>>> >>> interesting >>> >>>> approach. >>>> >>>> Cheers, >>>> >>>> Peter >>>> >>>> >>>> >>>> >>>> >>>> On Jun 1, 2009, at 15:13, Pankaj Jaiswal (OSU) wrote: >>>> >>>> >>>>> Barry Smith wrote: >>>>> >>>>>> At 01:45 PM 6/1/2009, Peter Midford wrote: >>>>>> >>>>>>> Barry, >>>>>>> >>>>>>>> Some comments >>>>>>>> >>>>>>>> It would be nice to see a definition of 'taxonomic rank', >>>>>>>> >>> since this >>> >>>>>>>> plays so overwhelming a role in the ontology. >>>>>>>> >>>>>>> Agree, >>>>>>> >>>>>>> Here's the current definition in the NCBI taxonomy ontology >>>>>>> "A level of depth within a species taxonomic tree" >>>>>>> >>>>>>> Definition in ICZN glossary (of rank) >>>>>>> "the level, for nomenclatural purposes, of a taxon in a taxonomic >>>>>>> hierarchy(e.g. all families are for nomenclatural purposes at the >>>>>>> same rank, which lies between superfamily and subfamily)." >>>>>>> >>>>>>> I would suggest combining these into: A level of depth of a >>>>>>> >>> taxon in >>> >>>>>>> a taxonomic hierarchy. This avoids the specification of a >>>>>>> >>> 'species >>> >>>>>>> taxonomic tree,' since a taxonomy need neither extend down to the >>>>>>> species level, nor stop there (NCBI has terms below the species >>>>>>> level already). >>>>>>> >>>>>> I agree with this; however, do all of your examples really fit >>>>>> this >>>>>> definition? do not some of them span multiple levels, for example? >>>>>> >>>>>>>> [Typedef] >>>>>>>> id: TAXRANK:has_rank >>>>>>>> name: has_rank >>>>>>>> alt_id: TAXRANK:0000033 >>>>>>>> comment: A metadata relation more or less equivalent to the >>>>>>>> >>> similar >>> >>>>>>>> relation in NCBI taxonomy. >>>>>>>> is_metadata_tag: true >>>>>>>> >>>>>>>> I would like to see a definition of this relation following >>>>>>>> >>> the rules >>> >>>>>>>> set out in >>>>>>>> >>>>>>>> >>> <http://genomebiology.com/2005/6/5/R46>http://genomebiology.com/2005/6/5/R46 >>> >>>>>>> NCBI currently has >>>>>>> >>>>>>> a metadata relation between a class and its taxonomic rank (e.g., >>>>>>> species, family) >>>>>>> >>>>>>> This doesn't seem sufficient to me. If we treating taxonomic >>>>>>> identifiers as classes, then to avoid metaclasses, we should at >>>>>>> least specify a root superclass for all these identifiers. This >>>>>>> root could be defined in this ontology, then any taxonomic >>>>>>> >>> ontology >>> >>>>>>> using this relation would specify the root specified here as the >>>>>>> parent of whatever they use for a root (e..g, Chordata). >>>>>>> >>>>>>> So perhaps "a metadata relation between a subclass of >>>>>>> TAXRANK:root >>>>>>> and a subclass of TAXRANK:taxonomic_rank, that specifies the >>>>>>> depth >>>>>>> of the former in the taxonomic tree" >>>>>>> >>>>>> relations normally do not specify; in general the subject of >>>>>> >>> the verb >>> >>>>>> 'to specify' should be a datum, I think -- so this definition >>>>>> needs >>>>>> rephrasing (but do we really need such an odd relation at all?) >>>>>> >>>>>>>> You have many assertions of the form >>>>>>>> >>>>>>>> tribe is_a taxonomic_rank >>>>>>>> >>>>>>>> The recommended definition of 'is_a' tells us that >>>>>>>> >>>>>>>> if A is_a B, then every instance of A is an instance of B. >>>>>>>> >>>>>>>> Thus you are asserting that for every x, >>>>>>>> >>>>>>>> if x instance_of tribe, then x instance_of taxonomic rank. >>>>>>>> >>>>>>>> This does not seem right. Does this mean that you have another >>>>>>>> definition of is_a? Do you mean by 'is_a' perhaps >>>>>>>> >>> 'instance_of'? If >>> >>>>>>>> so, then you should say so. Then, however, we have not an >>>>>>>> >>> ontology of >>> >>>>>>>> ranks but rather, like GAZ, a controlled vocabulary built upon >>>>>>>> ontological principles. This is because ontologies represent >>>>>>>> only >>>>>>>> types, and you are representing just one type, together with >>>>>>>> providing a list of instances. >>>>>>>> BS >>>>>>>> >>>>>>> Except that none of these terms (expect rank itself) have is_a >>>>>>> children - the only children they have are via the has_rank >>>>>>> relation, which as a metadata relation supports no inference at >>>>>>> all. Doing it this way seems to work, but the design decision >>>>>>> (making has_rank a metadata relation) was originally made by >>>>>>> Chris >>>>>>> (I believe). >>>>>>> >>>>>> I think we need to involve Chris in this discussion >>>>>> BS >>>>>> >>>>> >>>>> How about using the rank names as relationship types. If so then >>>>> >>> we do >>> >>>>> not have to create a separate ranking system with 'rank' classes >>>>> >>> that >>> >>>>> have taxonomic names as children. The relationship_types can have >>>>> rankings of their own. >>>>> >>>>> If I go by later we may be proposing something like >>>>> >>>>> [i] = is_a >>>>> >>>>> >>>>>> Oryzeae >>>>>> >>>>> --[i] oryza >>>>> ----[i] oryza sativa >>>>> taxonomic rank >>>>> --[i] tribe >>>>> ----[i] Oryzeae >>>>> --[i] genus >>>>> ----[i] oryza >>>>> --[i] species >>>>> ----[i] oryza sativa >>>>> >>>>> However if we go by former by using ranks as relationship types, >>>>> >>> then >>> >>>>> it is more clear inference as follows and no need to extend the >>>>> data >>>>> structure to populate the rank classes. >>>>> >>>>> >>>>> Oryzeae >>>>> --[is_genus_of] oryza >>>>> ----[is_species_of] Oryza sativa >>>>> >>>>> Also what about the interspecific terms that we have many in plants >>>>> (not all listed on NCBI) >>>>> genetic cross of two taxons, e.g. >>>>> name: Triticum aestivum x Secale >>>>> synonym: Secale cereale x Triticum >>>>> >>>>> >>> aestivumcerealehttp://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Info&lvl=3&keep=1&srchmode=1&unlock&id=49318 >>> <http://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?mode=Info&lvl=3&keep=1&srchmode=1&unlock&id=49318 >>> >>>>> Terms like these can have two parents (the two parent species) >>>>> >>> may be >>> >>>>> with a different relationship_type (e.g. is_intersection_of >>>>> /interspecific_cross_of) other than species used by NCBI. >>>>> >>>> Peter E. Midford >>>> Phenoscape Taxonomy Curator >>>> Pet...@gm... <mailto:Pet...@gm...> >>>> >>>> >>>> >>>> >>>> >>> -- >>> Pankaj Jaiswal >>> Assistant Professor >>> Dept. of Botany and Plant Pathology >>> 3082 Cordley Hall >>> Oregon State University >>> Corvallis, OR, 97331-2902, USA >>> >>> Ph.: +1-541-737-8471 >>> Fax: +1-541-737-3573 >>> Web: www.gramene.org <http://www.gramene.org> >>> www.plantontology.org <http://www.plantontology.org> >>> >>> >>> >>> ------------------------------------------------------------------------------ >>> OpenSolaris 2009.06 is a cutting edge operating system for >>> enterprises >>> looking to deploy the next generation of Solaris that includes the >>> latest >>> innovations from Sun and the OpenSource community. Download a >>> copy and >>> enjoy capabilities such as Networking, Storage and Virtualization. >>> Go to: http://p.sf.net/sfu/opensolaris-get >>> _______________________________________________ >>> Obo-discuss mailing list >>> Obo...@li... >>> <mailto:Obo...@li...> >>> https://lists.sourceforge.net/lists/listinfo/obo-discuss >>> >>> >>> >>> ------------------------------------------------------------------------ >>> >>> ------------------------------------------------------------------------------ >>> OpenSolaris 2009.06 is a cutting edge operating system for >>> enterprises >>> looking to deploy the next generation of Solaris that includes the >>> latest >>> innovations from Sun and the OpenSource community. Download a copy >>> and >>> enjoy capabilities such as Networking, Storage and Virtualization. >>> Go to: http://p.sf.net/sfu/opensolaris-get >>> >>> >>> ------------------------------------------------------------------------ >>> >>> _______________________________________________ >>> Obo-discuss mailing list >>> Obo...@li... >>> https://lists.sourceforge.net/lists/listinfo/obo-discuss >>> >> -- >> Pankaj Jaiswal >> Assistant Professor >> Dept. of Botany and Plant Pathology >> 3082 Cordley Hall >> Oregon State University >> Corvallis, OR, 97331-2902, USA >> >> Ph.: +1-541-737-8471 >> Fax: +1-541-737-3573 >> Web: www.gramene.org >> www.plantontology.org >> >> _______________________________________________ >> Phenoscape mailing list >> Phe...@ne... >> https://lists.nescent.org/mailman/listinfo/phenoscape >> > > -- Pankaj Jaiswal Assistant Professor Dept. of Botany and Plant Pathology 3082 Cordley Hall Oregon State University Corvallis, OR, 97331-2902, USA Ph.: +1-541-737-8471 Fax: +1-541-737-3573 Web: www.gramene.org www.plantontology.org |